Marketing Insight

Marketing Insight Course Reader ICFAI University, Hyderabad Summer 2008, Term II Amiya K. Basu Professor of Marketing, Syracuse University May 31, 2008

Contents 1 Introduction 1.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Alternative Marketing Management Philosophies . . . 1.3 Elements of Marketing Strategy . . . . . . . . . . . . 1.4 Appendix: Brand-Switching Matrix . . . . . . . . . . 1.5 Example: Mini-Survey on Cellular Telephone Service 2 Selected Topics on Consumer Decision Making 2.1 Stages in Consumer Decision Making Process . 2.2 Types of Consumer Decision Making . . . . . . 2.3 Evaluation of Alternatives . . . . . . . . . . . . 2.4 Maslow’s Hierarchy of Needs . . . . . . . . . . . 2.5 Business Customers . . . . . . . . . . . . . . . . 2.6 Nature of Business Markets . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . .

. . . .

. . . .

. . . .

. . . . .

. . . . .

. . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Gamble . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

5 Conjoint Analysis 5.1 Idea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Laptop Example . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Evaluation of a New Product Concept using Linear Interpolation . 5.4 Prediction of Market-Share . . . . . . . . . . . . . . . . . . . . . . 5.5 Example: Credit Card . . . . . . . . . . . . . . . . . . . . . . . . 5.6 Example: Personal Computer . . . . . . . . . . . . . . . . . . . . i

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

and Business Marketing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 Selected Topics on Market Segmentation and Targeting 3.1 Examples of Market Segmentation in Practice: Procter and 3.2 Meaning of Market Segmentation . . . . . . . . . . . . . . 3.3 Evolution of the Concept of Market Segmentation . . . . . 3.4 Possible Benefits of Market Segmentation . . . . . . . . . . 3.5 Criteria for Successful Segmentation . . . . . . . . . . . . . 3.6 Bases for Segmenting Consumer Markets . . . . . . . . . . 3.7 Example of Psychographic Segmentation: VALS . . . . . . 3.8 Strategies for Selecting Target Markets . . . . . . . . . . . 3.9 Product Positioning Strategies . . . . . . . . . . . . . . . . 3.10 Example of Segmentation using student data . . . . . . . . 4 Selected Topics on Marketing Research 4.1 Introduction . . . . . . . . . . . . . . . . 4.2 Basics of Sampling . . . . . . . . . . . . 4.3 Non-Response in Mail Surveys . . . . . . 4.4 Example: Focus Group Outline . . . . .

. . . . .

. . . .

. . . . . .

. . . . . . . . . .

. . . .

. . . . . .

. . . . . . . . . .

. . . .

. . . . . .

. . . . . . . . . .

. . . .

. . . . . .

. . . . . . . . . .

. . . .

. . . . . .

. . . . . . . . . .

. . . .

. . . . . .

. . . . .

1 1 2 3 9 11

. . . . . .

20 20 21 22 29 29 30

. . . . . . . . . .

35 35 36 37 38 38 39 40 42 44 47

. . . .

51 51 60 65 67

. . . . . .

68 68 69 70 74 74 77

5.7 5.8 5.9

Optional Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Additional Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix: Survey for Credit Card and PC . . . . . . . . . . . . . . . . . . . .

81 84 84

6 Selected Topics on Product Strategy 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Classification of Consumer Products . . . . . . . . . . . . . . . . . . . . . . . 6.3 Branding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4 Product Life Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5 Product features that affect rate of new product adoption . . . . . . . . . . . . 6.6 Strategic Planning and Boston Consulting Group (BCG) Market Growth/Market Share Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.7 Appendix: Practice Problem on BCG Market Growth/Market Share Matrix . 6.8 Appendix: Hypothetical Example of Experience Curve Effect . . . . . . . . . .

91 91 91 95 98 102 103 107 108

7 Selected Topics on Pricing Strategy 7.1 Introduction and Notations . . . . . . . . . . . . . . 7.2 Demand and Own Price Elasticity of Demand . . . . 7.3 Factors That Affect Price Sensitivity . . . . . . . . . 7.4 Price Quality Relationships . . . . . . . . . . . . . . 7.5 Basic Methods of Setting Price . . . . . . . . . . . . 7.6 Competitive Bidding . . . . . . . . . . . . . . . . . . 7.7 Psychological Pricing . . . . . . . . . . . . . . . . . . 7.8 Skim and Penetration Pricing . . . . . . . . . . . . . 7.9 Spatial (Geographic) Pricing . . . . . . . . . . . . . . 7.10 Cross Price Elasticity of Demand and Pricing Related 7.11 Additional Topics . . . . . . . . . . . . . . . . . . . . 7.12 Demand Functions used in Practice . . . . . . . . . . 7.13 Exercise Problems . . . . . . . . . . . . . . . . . . . . 7.14 Additional Reading . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

109 109 111 117 119 119 122 125 126 127 128 129 133 139 141

. . . . . . . . .

142 142 143 144 147 149 151 152 154 157

8 Selected Topics on Distribution Strategy 8.1 Rationale for Marketing Intermediaries (Middlemen) 8.2 Reasons for having intermediaries . . . . . . . . . . . 8.3 Distribution Intensity . . . . . . . . . . . . . . . . . . 8.4 Vertical and Horizontal Conflict . . . . . . . . . . . . 8.5 Alternative Channel Arrangements . . . . . . . . . . 8.6 Power in a Channel of Distribution . . . . . . . . . . 8.7 Some Legal Issues Related To Distribution Strategy . 8.8 Channel Coordination . . . . . . . . . . . . . . . . . 8.9 Additional Reading . . . . . . . . . . . . . . . . . . . ii

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Products . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . .

9 Selected Topics on Promotion Strategy 9.1 Basic Ideas . . . . . . . . . . . . . . . 9.2 Selected Topics on Personal Selling . . 9.3 Publicity . . . . . . . . . . . . . . . . . 9.4 Selected Topics on Advertising . . . . . 9.5 Sales Promotions . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

10 Estimation of Conjoint Model (Optional Material) 10.1 Dummy Variable Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Back to Conjoint Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 Estimation of Additive Part-Worth Model using Dummy Variable Regression 10.4 Estimation of Conjoint Model for Credit Card Data in Section 5.5 . . . . . . 10.5 Estimation of Conjoint Model for PC Data from Section 5.6 . . . . . . . . . 10.6 Estimation of Ideal Point Model . . . . . . . . . . . . . . . . . . . . . . . . . 11 Questionnaire Construction (Optional Material) 11.1 Basic Steps in Questionnaire Construction . . . . 11.2 Different Question Forms . . . . . . . . . . . . . . 11.3 Common Mistakes in Questionnaire Construction 11.4 Itemized Rating Scales . . . . . . . . . . . . . . . 11.5 Coding a Questionnaire . . . . . . . . . . . . . . . 11.6 Data Editing and Cleaning . . . . . . . . . . . . . 11.7 Example of Survey and Coding Scheme . . . . . .

iii

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . .

158 158 160 168 169 173

. . . . . .

177 177 183 184 185 187 189

. . . . . . .

191 191 192 195 198 200 203 206

1

Introduction

1.1

Basics

Marketing: Marketing is the activity, set of institutions, and processes for creating, communicating, delivering, and exchanging offerings that have value for customers, clients, partners, and society at large. (Adopted by the American Marketing Association in October, 2007) More simply, Marketing is a process of exchange between two entities, a marketer who produces a good, service or idea, and a customer. Product: Anything that one can receive in an exchange. Ideally, a product should be something that can satisfy a need or want. That may include physical goods (e.g., an Apple iPOD), services (e.g., Apple iTunes music download service), combination of goods and services (Apple iPOD with one year warranty), issues, and ideas. Exchange: Process by which two or more parties give something of value to each other to satisfy perceived needs. An exchange requires the following conditions: • At least two parties. • Each party has something that might be of value to the other party. • Each party is capable of communication and delivery. • Each party is free to accept or reject the exchange offer. • Each party feels it is appropriate or desirable to deal with the other party. What is involved in marketing? • Identification of customer needs. • Design of goods and services to meet those needs. • Promotion, that is, communication of information about those goods and services to potential customers (key elements: advertising, personal selling, sales promotion, public relations). • Distribution of the product (time, place). • Setting the price of the product. • Service and follow up.

1

1.2

Alternative Marketing Management Philosophies

1.2.1 The Production Concept: This philosophy focuses on the internal capabilities of the firm in terms of how efficiently it can produce and distribute the product. The idea is to streamline production and distribution to reduce cost. That allows the marketer to offer a low price and hence generate a large demand. This, in turn, should give the marketer economy of scale, and the process of cost reduction will be reinforced. 1.2.2 The Product Concept: Like the production orientation, the product orientation also focuses on the internal capabilities of the firm. The objective now is to achieve a product of high quality. The idea is that if product quality is high, customers would recognize that fact and buy the product. Potential Problems with Production and Product Concepts: • The production and product concepts are incomplete because they do not consider what the customer wants. You may have a great product and produce it efficiently, but nobody may want it. • While it is important to offer a quality product, it is also necessary to examine how customers make trade-offs between price and quality. Otherwise, the product may be too expensive to succeed in the marketplace. 1.2.3 The Selling Concept: This concept encourages the marketer to focus primarily on promoting the product to customers and marketing intermediaries (resellers). Some methods used are: • Heavy advertising targeted at the end customer. • Strong incentives to salespeople to sell the product. That may be done by using high commission rates, bonuses for exceeding sales quotas, and sales contests. • Promotional allowances (money to the reseller to promote the product), slotting allowances (fixed fee paid to the reseller just to carry the product), and co-operative advertising (partly reimburse the cost of local advertising) to the reseller. Potential Problems with the Selling Concept: 1. The customer may not be interested in the product offered. 2. Heavy incentives encourage salespeople to hard-sell the product. This can generate additional sales in the short runs, but customers may return the product later. Also, if the customer is pressured to buy a product, they may develop a negative attitude about the marketer. 1.2.4 The Marketing Concept: The idea is simple: The marketer should try to satisfy the customers wants and needs while meeting the organization’s objectives. This includes the following: 2

• Customer wants and needs come first. The organization must focus of customer wants and needs in order to distinguish its products from competitor’s offerings. • This must be a company-wide orientation. The organization should integrate all its activities (including production) to achieve the objective of satisfying customer wants and needs. • Achieve long term goals of the organization by satisfying customer wants and needs legally and responsibly. Issue: What happens when the product satisfies customer needs and wants legally, but the product may be harmful to the customer or broader society in the long run? 1.2.5 The Societal Marketing Concept: Achieve a balance between what the customer wants and what is good for the customer and society at large in the long run. Issue: Usually more costly to implement than the marketing concept. It can only be implemented if at least one of the following happens: • There are government regulations or incentives that encourage societal marketing (e.g., tax break for buying a hybrid car). • There is a substantial body of customers ready to pay extra for a product that is better for society at large (e.g., Fair Trade coffee from plantation that give workers a living wage). • There is a vigilant body of consumer activists that will organize public opinion against marketers who harm society in the long run. • The industry regulates itself. This is often done to preempt government regulation or consumer boycott.

1.3

Elements of Marketing Strategy

1.3.1 Basic Ideas: Target Market: Group of people or organizations for which a marketer designs, implements, and maintains a marketing mix. Objective: Satisfy their specific needs and wants, resulting in mutually satisfactory exchanges. Decision Variables of Marketer: Marketing Mix (also called the 4 P’s of Marketing) 1. Product 2. Price 3. Promotion: Advertising, personal selling, sales promotion, public relations 3

4. Distribution (place, time) Marketing Environment Micro Environment 1. Internal environment • Cost Structure • Workforce • Capabilities • Resources available 2. Suppliers 3. Marketing intermediaries 4. Customers 5. Competitors 6. Publics: Any group of people that has an actual or potential interest in or impact on the organization’s ability to achieve its goals. Clearly, the organization needs to communicate with its publics effectively. Typically they include media, shareholders, employees, neighbors, and regulators. Macro Environment 1. Demographic 2. Economic (for example, inflation, recession, interest rate, currency exchange rates) 3. Technological 4. Natural 5. Political/legal 6. Social/cultural 1.3.2 Examples of changes in Demographic Environment Example 1: Aging of baby-boomers Example 2: Women in the work-force Example 3: Rise of Generation Y. Born between 1979 and 1994, the Generation Y is a technologically sophisticated and internationally connected group of customers. At least two companies depended on Generation Y for success: Apple (iPod), and YouTube. Discussion Problem: Identify a marketing opportunity created by each of the above changes. 1.3.3 Technological environment: Some factors 4

1. Internet: The rise of the internet has had profound impact on may areas of marketing. Some examples: • It has allowed marketers serving a small and spatially dispersed group of customers to survive. For example, a marketer who sells packaged Turkish sweets can directly serve customers scattered over a large geographic area without depending on middlemen. • It has created a platform for people to buy and sell products (eBay, Amazon). • It allows marketers to distribute digital products (music, movie, online games, software) without any distortion and at a very low marginal cost. 2. Modular products: For many product categories, the end product is a combination of modules. For example, a personal computer offered by Dell is the combination of its motherboard, hard drive, CD/DVD drives, RAM, monitor, and different software packages that come with it. By changing the combination of modules, it is possible to customize a product at a low cost. At the same time, since each module is mass produced, economy of scale is achieved. 3. Supermarket scanners that read the bar codes on packages. This allows better record keeping and collecting information on consumer behavior through customer cards. Consequence: This has shifted power in channels of distribution from the manufacturer to the retailer, as retailers often have greater knowledge of demand than manufacturers. As a result, retailers now charge promotional allowances, slotting allowances. 1.3.4 Political/Legal Environment The political/legal environment affects many areas of business management. For example, in the USA, the changes in the federal minimum wage affects company hiring, payroll, and costs of production. The SarbanesOxley Act of 2002, passed in response to corporate and accounting scandals such as Enron and Tyco, has had a major effect on accounting practices and financial disclosure by companies. The areas listed below are only the areas of law that affect marketing decisions in the USA. Major Objectives of US Federal Laws Relating to Marketing: (a) To maintain competition and prevent the formation of monopolies. Examples: • Sherman Antitrust Act (1890): Makes trusts and conspiracies in restraint in trade illegal; makes monopolies and attempts to monopolize a misdemeanor. • Clayton Act (1914): Outlaws prices discrimination, tying agreements. • Federal Trade Commission (FTC) Act (1914) • Robinson-Patman Act (1936): Prohibits charging different prices to different buyers of merchandize of similar grade and quantity; requires sellers to offer allowances to all purchasers on proportionately equal basis. Applies only to situations where a marketer in a state in the USA sells its products to other states in the USA. 5

(b) To protect consumers from unfair business practices. These laws do not wish to prevent marketers from doing business. The goal of these laws is limited to consumer protection. Examples: • Meat Inspection Act • Automobile Information Disclosure Act • Fair Packaging and Labeling Act (1966) • Child Protection Act (1966): Bans sales of hazardous toys. Sets standards for childresistant packaging. • Nutrition Labeling and Education Act (1990): Requires that food product labels provide detailed nutritional information. (c) Protect Interests of society from unrestrained business behavior. The goal of these laws is to reduce or even eliminate certain marketing activities. Examples: • Federal Cigarette Labelling and Advertising Act (1967): Cigarette packages must include the Surgeon General’s warning. • Children’s Television Act (1990): Limits the number of commercials during children’s programs. (d) De-regulation. Examples: FTC Improvement Act, deregulation of the transportation industry. 1.3.5 Competitive Environment Key to understanding competitive environment: (1) Identification of the need being satisfied. If two products can satisfy the same need, then they are competing products. (2) A clear idea of how the customer chooses a product. For example, if the choice follows a series of steps, how is the choice made at each step? Primary and Selective Demand: Primary demand is the demand for a product category. Selective demand is the demand for a product within the category. How to identify your competitor? 1. Frequently purchased brands: Look at purchase data over many purchase periods. One method is examination of a “brand switching matrix” discussed later. 2. Durable products: Need to determine how close two objects are in the mind of a consumer (“perceptual space”). 6

3. Use a survey to find the “evoked set” of a customer. Explanation of Terms: 1. Perceptual space: A consumer perceives a product in terms of the product’s features (also called “attributes”). For example, a university may be perceived in terms of its tuition rate, location, and academic reputation. Thus, each product alternative can be expressed as a point in a multidimensional space in the mind of the customer. This space is called the perceptual space. 2. Evoked Set: The evoked set of a consumer is the set of products (s)he considers before making the choice. For example, Syracuse University might like to know which other schools a student had applied to. This can be used to identify the competition for both durable and frequently purchased products. Example: Identifying competition using purchase history You have data from 2000 consumers. Each customer in this sample bought exactly one tube of toothpaste in April 2006, and one tube of toothpaste in May 2006. Each customer purchased one of four brands: A, B, C, D. The data are presented as a cross-tab where the entry in the i j-th cell is the number of consumers in your sample who purchased brand i in April and j in May. For example, 200 out of this sample of 2000 purchased A in April, and B in May. May 2006 April 2006 A B C D Row Totals A 500 200 0 0 700 B 100 400 0 0 500 C 0 0 300 100 400 D 0 0 0 400 400 Column Totals 600 600 300 500 Total = 2000 Note: To understand what is going on, consider Row 1. The total of this row, i.e., 700, is how many people purchased brand A in month 1. Out of these 700 people, 500 purchased A again in month 2, 200 switched to B, and nobody switched to C or D. Question 1: Which product is competing with which? Question 2: Which product is in the greatest trouble? Answers: There are two competitive groups: (1) brand A is competing with brand B (60% of the overall market), and (2) brand C is competing with brand D (40% of the overall market). The nature of competition is different in the two groups. Brand A seems to be an incumbent product (such as Crest Fluoride), and B is a relative new comer (such as Colgate Fluoride). Consumers seem to find the two products close substitutes as there is switching back and forth between them. In the long run, both products will probably have equal shares of their portion of the market (60% of the overall market). In contrast, brand D is drawing customers out of C while C is not getting any customers from D, indicating that D is perceived as a superior product. If this goes on, C may lose all its customers to D. Thus, while brands A and C are 7

both losing customers, C is in much more serious trouble. Important Note: The table given above is an example of a brand switching matrix. Looking at a brand switching matrix, you can (1) identify who your competitors are, and (2) learn how you are doing against your competitors. To construct a brand switching matrix, you need record of purchases from the same set of customers over successive purchase occasions. Such data can be obtained by (1) looking at shopping-card records if the customers always goes to the same grocery-chain (such as Wegmans) and uses a card, or, (2) requiring the customer keep a diary of purchases. Effect of Competition on Marketing Decisions 1. Product Decisions • Bench-marking: Attribute by attribute, determine how your offering compares with the alternative that performs best on that attribute. • Product Positioning: Creating a position of your product in the perceptual space of the customer. This often requires comparing your product with competitor’s products. • Product Differentiation: Make your product different from competing products on one or more dimensions. Communicate these differences to the customer. • Branding: Develop a brand name that tells the customer what to expect from your product. 2. Pricing decisions: If the customer has more similar alternatives to choose from, she will be more price sensitive. This has important impact on how the firm should set prices. Basically: • Less competition allows you to set your price higher. • If your product is perceived as unique, there will be a residual demand even when price is high. On the other hand, if it is perceived as a commodity with many close substitutes, demand will fall to zero if your price is much higher than the price charged by your competitors. 3. Promotion decisions: A major focus of any promotion decision is to highlight your product positioning. If your offering has unique features, usually your promotion tries to stress the difference and charges a premium price (POD strategy). On the other hand, if your strength is low cost, you may stress that you offer the same benefits as the expensive alternative at a lower price (POP strategy). 4. Distribution decisions: You need to examine the distribution strategy of your competitor to determine what threat or opportunity it presents. For example: 8

• Are you ignoring the specific needs of some customers? For instance, are there many students who wish to take evening MBA classes? Are there customers who wish to do their grocery shopping on Sundays? Are there households who would like to have groceries delivered at home? • Is your competitor ignoring specific needs of some customers that you can fulfill?

1.4

Appendix: Brand-Switching Matrix

1.4.1 Cross-Tabulation A cross-tabulation is used to summarize data obtained from a sample. This method deals with two variables, each of which has a finite number of categories in the sample considered. For example, for a sample of Syracuse University students, we may record (1) gender (male or female) and (2) major (business, arts/science, engineering, other). We can place these students into 2 × 4, that is, 8 categories: (1) male, business, (2) female, business, etc., and count how many members of the sample belong to each of the 8 categories. The table we get this way is called a cross-tabulation or cross-tab of these two variables. A hypothetical cross-tabulation for these two variables is given below: Major Business Arts/Science Engineering Other

Gender Male Female 20 20 15 25 20 10 18 22

For example, in the sample of size 150, 20 are male business majors, 25 are female arts/science majors, etc. Cross-tabs can be used to understand the nature of the relationship between two variables. 1.4.2 Brand-Switching Matrix: When you have data on what brand a customer selected in two successive purchase occasions, you can prepare a brand-switching matrix. A brandswitching matrix is a cross-tabulation of brand selected the first time against brand selected the second time. By looking at this matrix, you can identify which brands are considered substitutes of one another by customers. Example to work with: Suppose you have a sample of 20 Syracuse University students who went to campus area bars last Friday and Saturday. Each day, each of the students went to exactly one bar, and they went to one of three bars: A, B, and C. You have the following data on which bars were visited: 9

Student # 1 2 3 4 5 6 7 8 9 10

Friday A C C A B C A B A B

Saturday A C C B A C A B A B

Student # 11 12 13 14 15 16 17 18 19 20

Friday C B A A B A C A A A

Saturday C A B A B A C A B A

Prepare a cross-tab of bar visited Friday against bar visited Saturday. [Note that there are 9 possible combinations here: AA (i.e., A Friday & A Saturday, AB (i.e., A Friday and B Saturday), AC, BA, BB, BC, CA, CB, and CC.] Proceed as follows. Note that in each purchase occasion, a customer selected one of three bars (A, B, or C). Thus, the brand-switching matrix has three rows and three columns. Prepare the template for the matrix as shown below. Saturday Friday A B C A B C Then, for each cell in the matrix, count how many times that combination has occurred in the sample and enter that number in the cell. For example, the combination AA has happened 7 times in the sample (student # 1, 7, 9, 14, 16, 18, and 20). Thus, you enter 7 in the cell AA. These are the students who went to bar A on Friday, and again on Saturday. Similarly, you enter 3 in cell AB (students 4, 13, 19). Complete the table. This is the brand-switching matrix.

10

1.5

Example: Mini-Survey on Cellular Telephone Service

1. Gender:

Male

Female

2. Place of residence (check one): Dorm

Fraternity/sorority

South Campus

Rental apratment/house

With parents

Other

3. Do you drive a motor vehicle while in Syracuse? Yes

No

4. Do you subscribe to a land-line (not cellular) telephone service? Yes

No

5. Do you subscribe to cellular telephone service? Yes (If yes, then continue with question 6.) No (If no, then go to question 13.) 6. Approximately how much money do you spend each month on cellular telephone service? (Check one) $20 or less/month

$21-$30/month

$31-$40/month

$41-$50/month

$51-$60/month

$61-$70/month

$71-$80/month

$81-$90/month

$91 or more/month

7. How important are the following to you when you evaluate a cellular telephone service provider? (Please circle)

(a) Monthly fee (b) Price of handset (c) Long distance rate (d) Reliability of connection (e) Features included (e.g., text messaging)

Not important at all 1 1 1 1 1

Very important 2 2 2 2 2

3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7

8. In the space below, list the names of all the cellular telephone service providers you can think of.

11

9. Which cellular telephone service provider do you subscribe to? (Check one. If more than one, then check the one you use primarily.) AT & T

Cingular

Cricket

Nextel

Sprint PCS T Mobile Verizon Other 10. Overall, how satisfied are you with your current (primary) cellular telephone service provider? (Please circle) Very Dissatisfied Very Satisfied 1 2 3 4 5 6 7 11. If you had the option to either stay with the current provider or switch to a different provider free of cost, which provider would you choose? (Check one) AT & T

Cingular

Cricket

Nextel

Sprint PCS T Mobile Verizon Other 12. How do you rate your current (primary) cellular telephone service provider on each of the following? (Please circle) Very Poor (a) Monthly fee (b) Price of handset (c) Long distance rate (d) Reliability of connection (e) Features included (e.g., text messaging) 13. During an average week, how many 0

1

2

1 1 1 1 1

Excellent 2 2 2 2 2

3 3 3 3 3

4 4 4 4 4

5 5 5 5 5

6 6 6 6 6

7 7 7 7 7

issues of the Daily Orange do you read? (Check one) 3

4

5

14. During Fall 2003 semester, how often did you attend an event at the Carrier Dome? Never 1-2 times 3-4 times

5 or more times

**Thank You for Your Help**

12

Results (Data Collected 1/13/’04)

(A) Importance Weights A.1 Mean Importance Weights of Attributes Variable X7a X7b X7c X7d X7e

N 65 65 65 65 65

Mean 5.9539 3.9692 5.3385 6.6000 5.0000

Std Dev 1.1240 1.6391 1.7435 0.7246 1.2248

Minimum 3.000 1.000 1.000 3.000 2.000

Maximum 7.000 7.000 7.000 7.000 7.000

A.2 Frequencies of Importance Weights X7a 3 4 5 6 7

X7b 1 2 3 4 5 6 7

X7c 1 2 3 4 5 6 7

Frequency 2 6 12 18 27

Frequency 5 9 9 18 13 6 5

Frequency 3 4 3 6 12 16 21

Percent 3.08 9.23 18.46 27.69 41.54

Cumulative Frequency 2 8 20 38 65

Cumulative Percent 3.08 12.31 30.77 58.46 100.00

Percent 7.69 13.85 13.85 27.69 20.00 9.23 7.69

Cumulative Frequency 5 14 23 41 54 60 65

Cumulative Percent 7.69 21.54 35.38 63.08 83.08 92.31 100.00

Percent 4.62 6.15 4.62 9.23 18.46 24.62 32.31

Cumulative Frequency 3 7 10 16 28 44 65

Cumulative Percent 4.62 10.77 15.38 24.62 43.08 67.69 100.00 13

X7d 3 5 6 7

X7e 2 3 4 5 6 7

Frequency 1 3 16 45

Frequency 2 4 16 21 14 8

Percent 1.54 4.62 24.62 69.23

Cumulative Frequency 1 4 20 65

Cumulative Percent 1.54 6.15 30.77 100.00

Percent 3.08 6.15 24.62 32.31 21.54 12.31

Cumulative Frequency 2 6 22 43 57 65

Cumulative Percent 3.08 9.23 33.85 66.15 87.69 100.00

(B) Top of the Mind Awareness of Providers (List all you can think of) AT&T 0 1 N

Count Percent 13 20.00 52 80.00 65

Cingular 0 1 N

Nextel 0 1 N

Count Percent 30 46.15 35 53.85 65

Sprint 0 1 N

Count Percent 9 13.85 56 86.15 65

Other 0 1 N

Count Percent 41 63.08 24 36.92 65

Verizon 0 1 N

Count Percent 1 1.54 64 98.46 65

Count Percent 12 18.46 53 81.54 65

TMobile 0 1 N

(C) Ratings of Current Provider on Attributes X12a 1 2 3 4 5 6 7

Frequency 1 1 2 12 30 13 6

Percent 1.54 1.54 3.08 18.46 46.15 20.00 9.23

Cumulative Frequency 1 2 4 16 46 59 65

Cumulative Percent 1.54 3.08 6.15 24.62 70.77 90.77 100.00 14

Cricket 0 1 N

Count Percent 46 70.77 19 29.23 65 Count Percent 14 21.54 51 78.46 65

Percent 4.62 1.54 12.31 27.69 23.08 21.54 9.23

Cumulative Frequency 3 4 12 30 45 59 65

Cumulative Percent 4.62 6.15 18.46 46.15 69.23 90.77 100.00

X12c 1 3 4 5 6 7

Frequency 1 2 6 9 19 28

Percent 1.54 3.08 9.23 13.85 29.23 43.08

Cumulative Frequency 1 3 9 18 37 65

Cumulative Percent 1.54 4.62 13.85 27.69 56.92 100.00

X12d 1 3 4 5 6 7

Cumulative Frequency Percent Frequency 2 3.08 2 3 4.62 5 4 6.15 9 12 18.46 21 15 23.08 36 29 44.62 65

Cumulative Percent 3.08 7.69 13.85 32.31 55.38 100.00

Cumulative Frequency Percent Frequency 2 3.13 2 2 3.13 4 5 7.81 9 22 34.38 31 22 34.38 53 11 17.19 64 Frequency Missing = 1

Cumulative Percent 3.13 6.25 14.06 48.44 82.81 100.00

X12b 1 2 3 4 5 6 7

X12e 2 3 4 5 6 7

Frequency 3 1 8 18 15 14 6

(D) Attribute Ratings of Current Provider D.1 Aggregate Attribute Ratings 15

Variable X12a X12b X12c X12d X12e

N 65 65 65 65 64

Mean 5.0308 4.6462 5.9385 5.8462 5.4531

Std Dev 1.1315 1.4624 1.2733 1.4387 1.1537

Minimum 1.000 1.000 1.000 1.000 2.000

Maximum 7.000 7.000 7.000 7.000 7.000

D.2 Ratings on Key Attributes (a and d) Provider AT&T Wireless Cingular Nextel Sprint PCS T Mobile Verizon Wireless

Number 1 4 3 5 3 49

Mean of X12a 5.00 4.75 4.00 4.60 4.67 5.18

Mean of X12d 6.00 3.75 5.67 4.60 3.00 6.33

Mean of X10 (Satisfaction) 5.00 4.25 6.00 5.00 4.00 6.14

D.3 Ratings on Other Attributes (b, c, e) Provider

Number

AT&T Wireless Cingular Nextel Sprint PCS T Mobile Verizon Wireless

1 4 3 5 3 49

Mean of X12b 4.00 3.50 4.33 5.20 4.00 4.76

Mean of X12c 6.00 4.50 5.67 6.60 6.67 5.96

Mean of X12e 4.75 5.67 5.60 5.67 5.47

Meaning of Attributes: a Monthly fee

b Price of handset

d Reliability

e Features included

c Long distance rate

(E) Satisfaction with Current Provider (X10 ) Frequencies of X10 for aggregate sample: X10 1 2 3 4 5 6 7

Frequency 1 2 1 1 13 28 19

Percent 1.54 3.08 1.54 1.54 20.00 43.08 29.23

Cumulative Frequency 1 3 4 5 18 46 65

Cumulative Percent 1.54 4.62 6.15 7.69 27.69 70.77 100.00 16

Mean of X10 (Satisfaction) 5.00 4.25 6.00 5.00 4.00 6.14

Mean satisfaction for aggregate sample: N Mean 65 5.8153846

Std Dev 1.2486531

Minimum 1.0000000

Maximum 7.0000000

Mean Satisfaction for different providers Provider X9 = 1 (AT&T) X9 = 2 (Cingular) X9 = 4 (Nextel) X9 = 5 (Sprint PCS) X9 = 6 (T Mobile) X9 = 7 (Verizon)

N 1 4 3 5 3 49

Mean 5.00 4.25 6.00 5.00 4.00 6.14

Std Dev 1.5 0 1.73 2.65 0.89

Minimum 5.00 2.00 6.00 2.00 1.00 3.00

Maximum 5.00 5.00 6.00 6.00 6.00 7.00

(F) Market Shares of providers F.1 Actual Shares: X9 1 2 4 5 6 7

Frequency 1 4 3 5 3 49

Percent 1.54 6.15 4.62 7.69 4.62 75.38

F.2 Wish Shares: X11 1 2 3 4 5 6 7

Frequency 3 3 1 10 6 2 40

Percent 4.62 4.62 1.54 15.38 9.23 3.08 61.54

Coding Scheme for X9 and X11 : 1 AT&T 5 Sprint PCS

2 Cingular 3 Cricket 4 Nextel 6 T Mobile 7 Verizon

F.3 Cross-tabulation: Current Provider by Provider They Wish to have 17

X9 1 2 4 5 6 7 Columns Totals

1 0 0 0 1 0 2 3

2 0 1 0 0 0 2 3

X11 4 0 0 3 0 1 6 10

3 0 0 0 0 0 1 1

5 0 0 0 3 0 3 6

6 0 0 0 0 0 2 2

7 1 3 0 1 2 33 40

Row Totals 1 4 3 5 3 49 Sample Size = 65

(G) Usage of Cellular Service Results on $ spent (X6 : 1 if $20 or less, 2 if $21-$30, 3 if $31-$40, 4 if $41-$50, 5 if $51-$60, 6 if $61-$70, 7 if $71-$80, 8 if $81-$90, 9 if $91 or more) Frequencies:

X6 1 2 3 4 5 6 7 8 9

Frequency 1 2 11 13 24 5 2 1 6

Percent 1.54 3.08 16.92 20.00 36.92 7.69 3.08 1.54 9.23

Cumulative Frequency 1 3 14 27 51 56 58 59 65

Cumulative Percent 1.54 4.62 21.54 41.54 78.46 86.15 89.23 90.77 100.00

Means and Standard Deviations of X6 Provider AT&T Cingular Nextel Sprint PCS T Mobile Verizon Aggregate Sample

N 1 4 3 5 3 49 65

Mean 5.00 3.00 5.00 5.60 7.00 4.80 4.86

Std Dev 1.41 1.00 2.07 2.00 1.73 1.80

(H) Media Habits: Attendance of Dome Events H.1 Frequencies of X14 : 18

X14 1 (Never) 2 (1-2 times/semester) 3 (3-4 times/semester) 4 (5 or more times/semester)

Frequency 7 26 7 25

H.2 Cross-tabulation: Gender (X1) against Attendance of Dome events (X14): X1 0 (Women) 1 (Men) Column Totals

1 (Never) 4 3 7

2 (1-2/semester) 19 7 26

X14 3 (3-4/semester) 5 2 7

(5 or more/semester) 7 18 25

Row Totals 35 30 N=65

Discussion Questions: 1. Identify a company doing well, and a company not doing well. Compare them attribute by attribute. What advice would you give to the company not doing well? 2. Consider the company with the dominant market share. Does it face any competitive threat? Which competitor presents that threat? Why do you say so?

19

2

Selected Topics on Consumer Decision Making and Business Marketing

2.1

Stages in Consumer Decision Making Process

In making a purchase, a consumer may go through the following five stages: (1) Need Recognition: This may be something that develops over time, such as the need to have an MBA degree to achieve your professional goals, or an impulse (no prior planning) like seeing a box of candy at the supermarket and feeling hungry. (2) Information Search: A consumer may search information to learn more about the product category, and about the brands available. The search may be: • Internal: Here, the customer is trying to retrieve information from her own memory. For instance, when a person selects a vacation destination, he may remember places he visited as a child and liked, and decide to go there again. • External: These can be of two types: – Marketing-Controlled: Sometimes, the customer obtains information from sources controlled by the marketer. For example, a prospective MBA student may visit the web-sites of different business schools to learn about their MBA programs. Marketers generally try to show their own products favorably, and customers know this fact. – Non-marketing Controlled: Here, the customer is trying to obtain information from sources she feels are not controlled by marketers. For instance, before applying to a college, one may talk to friends and family members to learn more about alternative choices. Customers consider these information bias-free. Other examples of non-marketing-controlled information sources include the Consumer Reports, and www.tripadvisor.com. The amount of external search conducted by a consumer depends on: • The consumer’s interest in the product category. • The consumer’s knowledge about the product category, and alternatives available. • The consumer’s experience with the product category. • The consumer’s confidence in making a decision. After the information search, the consumer usually makes a short-list of alternatives to choose from. This is called the evoked set (also called “consideration set”) of the consumer. 20

Note: Marketers sometimes plant paid endorsers among communities of prospective customers. The P&G Tremor program is a major example of this trend. (3) Evaluation of Alternatives: We discuss this in greater detail later (see Section 2.3 of the Reader). (4) Purchase (5) Post-Purchase Behavior: During this stage, the consumer compares what she expected to get, and what she actually received. Any imbalance between what she expected and what she actually got after purchase is called cognitive dissonance. Cognitive dissonance can only occur after a purchase is made. It is likely to increase if: • The purchase is major. • It is hard to reverse the purchase act. • The rejected alternatives have desirable features the chosen alternative does not have. A consumer may try to reduce cognitive dissonance by seeking favorable information about the chosen alternative, and avoiding favorable information about alternatives not selected. If she still feels dissatisfied, she may decide to switch to another product in the future. Marketers recognize that cognitive dissonance is both a problem and an opportunity. The incumbent marketer tries to reduce cognitive dissonance by allowing returns, offering aftersales services, and sometimes running advertising campaigns trying to reassure the customer. Competing marketers try to increase cognitive dissonance by planting doubts in the mind of the customer about the purchase made. This is done through comparative advertising focusing on some weakness of the competitor (e.g., commercials by Hertz car rental agency, and the Capital One credit card). Note: While the five-stage framework discussed above is comprehensive, a consumer may not always go through all these stages. For example, during impulse purchase, a consumer may go straight from need recognition (seeing a box of candy and feeling hungry) to purchase.

2.2

Types of Consumer Decision Making

To study consumer decision making, it is useful to categorize different types of consumer decision making. There are two alternative ways we can do that. Categorization 1: Consumer decision making may be placed in three categories based on two dimensions: the consumer’s familiarity with the product category, and her familiarity with the brands available to choose from. The categories are: (1) Routine response behavior: Here, the consumer is familiar with both the product category, and the brands available to choose from. In such cases, the consumer often selects the brand she usually buys. For example, if you need laundry detergent, you probably buy the same brand each time. 21

(2) Limited decision making: Here, the consumer is familiar with the product category, but not familiar with the brands available. For example, if you are traveling in Europe and need to buy laundry detergent, you may need to learn about the laundry detergents available there. (3) Extensive decision making: Here, the consumer is unfamiliar with both the product category and the brands available and needs to study both of these to make a choice. Categorization 2: Alternatively, consumer decision making may be categorized based on two dimensions: • The consumer’s involvement with the purchase, which can be high or low. • (2) Differences among the alternatives as perceived by the consumer, which can be high or low. There are four possibilities: 1. High involvement, high perceived difference: complex buying behavior. Here, the consumer is highly involved with the purchase, and also is knowledgeable about the product category and the alternatives available. The consumer is likely to develop evaluative criteria based on the product attributes before making a final decision. 2. High involvement, low perceived difference: dissonance-reducing buying behavior. Here, the consumer is highly involved with the purchase, but does not have the knowledge or confidence to compare the product alternatives. In such cases, she is likely to play it safe and try to make sure she is not sorry after the purchase. Thus, the consumer is trying avoid cognitive dissonance. In these cases, factors such as return policy, the length of warranty, expert opinion, and prior experience with the product category are very important. 3. Low involvement, high perceived difference: variety-seeking behavior. Here, the consumer is likely to try out different alternative products. For example, you may buy different flavors of ice-cream on successive trips to the store. 4. Low involvement, low perceived difference: Habitual buying behavior. Here, the consumer does not see any need to try out different alternatives (they are all the same to her) and usually buys the same brand repeatedly.

2.3

Evaluation of Alternatives

Our discussion centers on limited or extensive decision making. (In routine response behavior, the consumer typically buys the same brand without comparing alternatives.) We assume that a consumer looks at each product alternative as a “profile” of its attribute values. For example, suppose a prospective MBA student has received offers of admission from four MBA programs, and she has rated each program in terms of the following attributes: reputation (X1 ), cost (X2 ), location (X3 ), and campus life (X4 ). For simplicity, suppose she rates each program on 22

a given attribute such as reputation on a 1-7 scale, from “very poor” (1) to excellent (7), and the rating scores are as follows: Program A B C D

Reputation (X1 ) 7 5 7 4

Cost (X2 ) 1 5 6 6

Location (X3 ) 6 5 3 6

Campus Life (X4 ) 3 5 6 6

We can think of a four dimensional space, the four attributes being the coordinates, where the decision makers perceives the four “products” (that is, MBA programs). This is the “perceptual space” here. (It is easy to visualize a perceptual space in two or three dimensions. Beyond three dimensions, it is harder to visualize a perceptual space, but the idea is the same as for lower dimensions.) Alternative Decision Rules: Once a consumer comes up with a perceptual space, she has to choose one alternative. No one knows for sure how a consumer makes a choice. Five basic rules which seem reasonable are described below. Note that in real life, a decision maker often uses a combination of one or more “basic” rules. Rule 1. Expectancy Value Rule: Here, we assume that when the consumer evaluates product alternatives, she assigns an “expectancy value” to each alternative, and selects the product with the highest expectancy value. She computes the expectancy values as follows: 1. She assigns an importance weight to every relevant attribute. For n attributes, we denote the importance weights by W1 , W2 , . . . , Wn . 2) For product alternative i under consideration with attribute values X1i , X2i , . . . , Xni , she computes the expectancy value (EMi ) as the weighted sum: EVi

=

W1 ∗ X1i + W2 ∗ X2i + . . . + Wn ∗ Xni .

This is repeated for every product alternative under consideration (W -s are same for all products, but X-s change from alternative to alternative). 3) She selects the alternative with the highest expectancy value. Note 1: The expectancy value rule is called a “compensatory” rule because a poor rating on one attribute may be balanced out (compensated for) by a good rating on another attribute. A poor rating on just one attribute may not necessarily imply that an alternative is rejected. Note 2: The choice of importance weights for a consumer is not unique. If all the importance weights (W -s) are multiplied by the same positive number K, all the expectancy values are also multiplied by that number K, and the choice remains the same. 23

For example, suppose there are four attributes (n = 4), and the respondent considers each attribute equally important. Then, for the purpose of selecting a product alternative, the following two alternative sets of importance weights will give the same result: • Alternative 1: W1 = 1, W2 = 1, W3 = 1, W4 = 1. Here, the expectancy value is the sum of individual attribute ratings. • Alternative 2: W1 = 0.25, W2 = 0.25, W3 = 0.25, W4 = 0.25 Example: Suppose the decision maker considers cost, location, and campus life equally important, and reputation twice as important as any one of the other three attributes. We can choose W1 = 2, W2 = 1, W3 = 1, and W4 = 1. Program A B C D

Expectancy Value (2 × 7) + (1 × 1) + (1 × 6) + (1 × 3) = 14 + 1 + 6 + 3 (2 × 5) + (1 × 5) + (1 × 5) + (1 × 5) = 10 + 5 + 5 + 5 (2 × 7) + (1 × 6) + (1 × 3) + (1 × 6) = 14 + 6 + 3 + 6 (2 × 4) + (1 × 6) + (1 × 6) + (1 × 6) = 8 + 6 + 6 + 6

= 24 = 25 = 29 = 26

Choice: Alternative C, with the highest expectancy value of 29, will be chosen. Note that if we multiplied by each W by the same positive number, choice would remain the same. In the present case, suppose we chose W1 = 0.4, W2 = W3 = W4 = 0.2. Then: Program A B C D

Expectancy Value (0.4 × 7) + (0.2 × 1) + (0.2 × 6) + (0.2 × 3) = 2.8 + 0.2 + 1.2 + 0.6 = 4.8 (0.4 × 5) + (0.2 × 5) + (0.2 × 5) + (0.2 × 5) = 2 + 1 + 1 + 1 = 5.0 (0.4 × 7) + (0.2 × 6) + (0.2 × 3) + (0.2 × 6) = 2.8 + 1.2 + 0.6 + 1.2 = 5.8 (0.4 × 4) + (0.2 × 6) + (0.2 × 6) + (0.2 × 6) = 1.6 + 1.2 + 1.2 + 1.2 = 5.2

Once again, the choice is Alternative C, with the highest expectancy value of 5.8. Rule 2. Conjunctive Rule: This rule is often used by a consumer to make a short list of alternatives for further consideration. According to this rule, the respondent sets minimum acceptable levels on one or more attributes. A product is considered further only if it meets all these criteria. Example: Consider the prospective MBA student introduced earlier. Suppose she only considers MBA programs that rate 6 or more on X1 , and 6 or more on X4 . Only MBA program C meets both criteria and is selected. Conjunctive rules often arise when a risk averse consumer is trying to make sure the product meets minimum standards on one or more dimensions. That happens in dissonance reducing buying behavior. The Kotler/Keller text gives a special case of the conjunctive rule where the customer selects the first product that satisfy all the criteria. More generally, a conjunctive rule is used to make a short-list from which the consumer makes a final choice. We use the more general definition in our class. 24

Rule 3. Disjunctive Rule: This rule is also used to prepare a short list of alternatives for further consideration. Here, the respondent sets minimum acceptable levels on one or more attributes. An alternative is considered further if it meets any one of these criteria. Example: Consider the prospective MBA student introduced earlier. Suppose she considers any MBA program that rates 6 or more on X1 , or 6 or more on X4 . MBA programs A, C, and D all meet the condition that at least one of the two criteria is satisfied: • A meets the first criterion. • C meets both criteria. • D meets the second criterion. An MBA program is only removed from consideration if it fails to meet both criteria. Disjunctive rules often arise in variety seeking buying behavior where the consumer is open to trying many alternatives as long as they show promise on some dimension. Rule 4. Lexicographic Rule: A consumer using this rule arranges the attributes from most important to least important. She then compares the alternatives as follows: 1. Start with the most important attribute. 2. Compare all the alternatives on this attribute alone. If any alternative is worse than any other on this attribute, it is eliminated from further consideration. If one alternative is better than all others on this attribute, this alternative is selected, and the search stops. If there is a tie on this attribute, keep only the set of alternatives that are tied for highest score on this attribute, and proceed to the next most important attribute. 3. Repeat step 2 until there is one alternative left. Example: Consider again our prospective MBA student. Suppose she uses a lexicographic decision rule where X1 is most important, X4 is second most important, X2 is third most important, and X3 is fourth most important. Her choice process will proceed as follows: Stage 1 2 3 4

Attribute Alternatives Considered X1 A, B, C, D X4 A, C X2 X3

Eliminated B, D A

Here, MBA program C is selected in stage 2, and the choice process stops there. Note: The name “lexicographic” comes from how the words in a dictionary (lexicon) are arranged. The key aspect of the lexicographic rule is that if an alternative is even slightly 25

worse on a more important attribute, then it is selected even if it has much superior ratings on less important attributes. Thus, the lexicographic rule is not “compensatory.” It should be relatively rare when a decision maker uses a lexicographic rule. It may be reasonable, for instance, when one is selecting a safety product such as baby car seat. Then, even when a brand is just slightly more safe, it may be chosen regardless of price. 5. Elimination by Aspect (EBA): In contrast to the rules discussed so far, EBA is a probabilistic decision rule. Here, the customer makes a list of desirable aspects a product alternative may have, and assigns a value to each aspect. For instance, suppose the prospective MBA student in our example has listed the following aspects and values: Aspect A1 A2 A3

Description 6 or more on reputation 6 or more on cost 6 or more on location

Value 40 40 20

The choice proceeds in stages. At each stage, the customer randomly chooses an aspect from the aspects not already chosen. The probability that an aspect is selected is proportional to the value of the aspect, that is, equal to the value of the aspect divided by the sum of the values of all the aspects not selected in a previous stage. After an aspect is chosen, two things can happen. If only one alternative possesses the aspect, that alternative is chosen. If two or more alternatives possess the aspect, then only the tied alternatives are considered further, and another aspect is chosen from the aspects not selected in an earlier stage. In the present example, the alternatives do or do not have aspects as shown below: Alternative A B C D

A1 Yes No Yes No

A2 No No Yes Yes

A3 Yes No No Yes

We have the following possibilities: Stage 1 A1 selected

Alternatives A, B, C, D

Eliminated B, D

A2 selected

A, B, C, D

A, B

A3 selected

A, B, C, D

B, C

Stage 2 A2 selected A3 selected A1 selected A3 selected A1 selected A2 selected

Alternatives A, C A, C C, D C, D A, D A, D

Eliminated A C D C D A

Choice C A C D A D

Program A: Program A may be selected in two ways: (1) Aspect 1 is chosen in stage 1, and then Aspect 3 is selected in stage 2. The probability 26

that this happens is: P (1 chosen in stage 1) ∗ P (3 chosen in stage 2|1 chosen in stage 1) =

³

´ ³ 20 ´ 40 ∗ = 0.4 × 0.333 = 0.1332 40 + 40 + 20 40 + 20

(2) Aspect 3 is chosen in Stage 1, and then Aspect 1 in selected in Stage 2. The probability that this happens is: P (3 chosen in stage 1) ∗ P (1 chosen in stage 2|3 chosen in stage 1) ³

´ ³ 40 ´ 20 ∗ = 0.2 × 0.5 = 0.10 40 + 40 + 20 40 + 40 Thus, the probability that Program A is chosen is 0.1332 + 0.10 = 0.2332.

=

Program B: Program B has none of the three aspects and is never selected. Program C: Program C may be selected in the following ways: (1) Aspect 1 in selected in stage 1, and Aspect 2 is selected in stage 2. The probability that this happens is: P (1 chosen in stage 1) ∗ P (2 chosen in stage 2|1 chosen in stage 1) =

³

´ ³ 40 ´ 40 ∗ = 0.4 × 0.667 = 0.2668 40 + 40 + 20 40 + 20

(2) Aspect 2 is selected in stage 1, and Aspect 1 is selected in stage 2. The probability that this happens is: P (2 chosen in stage 1) ∗ P (1 chosen in stage 2|2 chosen in stage 1) ³

´ ³ 40 ´ 40 ∗ = 0.4 × 0.667 = 0.2668 40 + 40 + 20 40 + 20 Thus, the probability Program C is chosen is 0.2668 + 0.2668 = 0.5336.

=

Program D: Program D may be chosen in the following ways: (1) Aspect 2 is selected in stage 1, and Aspect 3 is selected in stage 2. The probability that this happens is: P (2 chosen in stage 1) ∗ P (3 chosen in stage 2|2 chosen in stage 1) =

³

´ ³ 20 ´ 40 ∗ = 0.4 × 0.333 = 0.1332 40 + 40 + 20 40 + 20

(2) Aspect 3 is selected in stage 1, and Aspect 2 is selected in stage 2. The probability that this happens is: P (3 chosen in stage 1) ∗ P (2 chosen in stage 2|3 chosen in stage 1) 27

=

³

´ ³ 40 ´ 20 ∗ = 0.2 × 0.5 = 0.10 40 + 40 + 20 40 + 40

Thus, the probability D is chosen is 0.1332 + 0.10 = 0.2332 General Notes: • If there is only a single cut-off point, then there is no difference between a conjunctive and a disjunctive rule. The rules become different if there are two or more cut-off points. Then, the conjunctive rule requires that all the cut-offs must be satisfied for the product to be considered further, while the disjunctive rule only requires that any one of the cut-offs be satisfied. • The lexicographic, conjunctive, disjunctive and EBA rules are not compensatory rules as a high level of one attribute may not be able to compensate for low levels on other attributes. For example, a consumer who requires a minimum level of gas mileage may not buy a car even though it offers excellent acceleration, handling and low price. • The rules described above are basic decision rules. In reality, a decision maker may use more than one of them to choose a product. For example, a consumer may make his selection in two stages: first use a conjunctive or a disjunctive rule to obtain a set of acceptable alternatives, and then use an expectancy value rule to make the final choice. For example, you may have used a conjunctive rule to eliminate schools outside the northeastern USA from further consideration, and then selected Syracuse from the resulting smaller set of alternatives using an expectancy value rule. • There is a simple extension of the expectancy value model where it is assumed that every consumer has an “ideal product” she compares all alternatives with. The “ideal product” occupies the “ideal point” in the “perceptual space” of the consumer, and she chooses the product which has the smallest distance from the “ideal point” in the consumer’s perceptual space. This model, called the “ideal point” model, is a compensatory model, that is, a low rating on one attribute can be balanced by a high rating on another attribute. We use this idea in Assignment 2. • For each rule described above, we may have a tie among two or more alternatives. For example, two MBA programs may have the same expectancy value. The rule itself does not tell you how the tie is broken. • For the conjunctive and disjunctive rules, it is possible that no alternative meets the requirement for selection. • For the elimination by aspect rule, it is possible that in a given stage, no alternative has the aspect selected. If that happens, we will ignore that aspect and proceed to select an aspect from the remaining aspects. 28

2.4

Maslow’s Hierarchy of Needs

A marketer is always interested in learning how important different attributes are to a consumer. However, attributes are product specific, and it is very difficult to determine what attributes of a new product will even be relevant to a consumer. Maslow’s hierarchy of needs is an attempt to understand what is more important to a consumer in general. According to this hierarchy, a consumer has the following five levels of needs, ranging from the lowest (1: physiological) to the highest (5: self actualization): 1. Physiological: Food, drink, shelter. Basically, survival. 2. Safety: security, protection, order. For example, you may buy a fire-wall to secure your computer from identity theft. 3. Social: affection, acceptance, belonging to a group. For example, one may buy perfume to be more attractive to others. 4. Esteem: respect, reputation, prestige, status. For example, one may purchase an expensive car to show his status in society. 5. Self Actualization: self fulfillment. For example, one may come back to school after retirement to get a graduate degree. The basic idea of Maslow’s hierarchy is that a lower need is always more important than a higher need, and a consumer tries to fulfill a higher need only when the lower needs are largely fulfilled. If a marketer has a product that satisfies a higher need and a lower need, Maslow’s hierarchy suggests that the product should be positioned by the lower need.

2.5

Business Customers

(a) Producers: They include • Manufacturing • Construction • Transportation • Services (legal, finance, food, security, health management) • Farm (b) Resellers: Buy finished goods and resell for profit. They include distributors (wholesalers who resell to business customers), and retailers (who resell to final consumers). (c) Governments: Federal, local, state, foreign (d) Institutions: Hospitals, universities, etc. These tend to be non-profit organizations. 29

2.6

Nature of Business Markets

2.6.1 Differences from Consumer Markets: 1. Market structure: Fewer but generally larger buyers. 2. Demand: derived, inelastic, fluctuates more. 3. Nature of buying unit: More buyers involved, more professional purchasing. 4. Types of decisions and decision process: more complex decisions, more formal buying process, long term relationship between buyer and seller possible. 5. Other: shorter channels, reciprocity, leasing more common. In general, compared to the consumer market, decision making in the business market is more complex and likely to involve more participants. Also, decision makers are usually more informed customers than the typical consumer. Because each decision is time consuming, purchase size is usually larger, and new purchase decisions are less frequent. Seller’s perspective: Business buyers often demand more information than the consumer, and the final product needs greater customization. As a result, business marketing involves greater emphasis on personal selling than advertising. Also, because of the larger size of a purchase order, there is less need for market intermediaries and channels of distribution are shorter. Sellers sometimes use national account management where a sales team is dedicated to one major client with divisions dispersed nationally. 2.6.2 Types of Business Products: (1) Major Equipment (Installations): major items like buildings, aircraft, computer systems, etc. Represents a major purchase, and is often customized to the needs of a specific customer. (2) Accessory Equipment: smaller items like personal computers, cars, etc., essential for running the business. (3) Component parts (modules such as computer chip, hard drive, DVD ROM, muffler for a car), processed materials (e.g., alloy steel, silicon wafer), and raw materials (e.g., corn to make cereals from). (4) Supplies: maintenance, repair, operating. (5) Services: financial, legal, medical, delivery, transportation, security, etc. 2.6.3 Type of Business Markets: (1) Vertical Market: Only a specialized set of customers buy your product. For example, Intel Pentium chips are sold only to PC manufacturers, and airplane navigational equipments are sold only manufacturers and repairing facilities of airplanes. Here, product knowledge is often very important, and the sales volume to a given customer is often large. Thus, channels of distribution tend to be shorter and there is much greater emphasis on personal selling (which can be more precisely targeted and can convey more information) than on advertising. (However, you do see TV commercials for Intel even though its customers are business customers. Why?) 30

(2) Horizontal Market: A wide spectrum of customers in different types of organizations may buy your product. Examples: markets for (i) personal computers, (ii) air conditioners, (iii) copy machines, (iv) Hammermill paper. Compared to vertical markets, here the marketer does more advertising and also do more selling through wholesalers, retailers, etc. 2.6.4 Characteristics of Demand in Business Markets: 1. Derived Demand: Demand for a business product is usually derived from demand in the consumer market. For example, there is a demand for Intel Pentium chips because consumers buy personal computers, that is, the demand for Pentium chips is derived from the demand for PC’s in the consumer market. Consequences of derived demand: 1(a). Inelastic demand: Quite often, a business product represents only a part of the final consumer product. Therefore, if the price of the business product changes, that has very little effect on the total price the consumer pays and thus does not change demand in the consumer market. Thus the demand for the business product often does not change significantly with price, that is, demand tends to be inelastic. For example, suppose the final product is a notebook computer which sells at a retail price of $1500. Consider a component of the computer, say, a graphics chip, which costs the manufacturer $100/unit. If the price of the chip increases to $110 and the manufacturer passes on the cost to the end customer, then the retail price of the notebook is $1510, less than a 1% increase. That is unlikely to have a major impact on the demand of the notebook computer and hence of the graphics chip. 1(b). Highly fluctuating demand: Consider, for example, demand for installations such as factories. Businesses like to operate at or near full capacity. Thus, even if demand in the consumer market drops slightly, factories are closed in order to consolidate production. On the other hand, if demand in the consumer market increases slightly, new factories are opened. Thus, slight changes in consumer demand can cause major changes in demand for business products. (Exception: Farm products usually do not have highly fluctuating demand because demand for food is stable.) 2. Joint demand: If two or more business products are all used to make the same consumer product, they have joint demand. For example, computer chips, LCD screens, hard drives, and DVD ROM’s are all used to make PC’s and hence their demands are joint. Consequently, a change in demand for the consumer product affect demands for all these products the same way. Thus, a change in demand at the consumer level can affect demand in a wide spectrum of industries. Example: Processor, LCD screen, and DVD ROM are all components of a note-book computer. Thus, the demands for all these components are derived from the demand for notebooks. If the demand for any component rises or falls, the demands for the other components do that also. The demands for the components are joint. 31

Processor

Joint ¾Demand

-

Joint Demand ¾

LCD

DVD

-

Screen

ROM

@

@

@

Derived

@

¡

Demand

¡

@

@

¡ ¡

¡

@

¡

@ R

?

¡

ª ¡

Notebook Computer

?

Consumer

2.6.5 Classifying Business Buying Behavior: In any business buying situation, a customer must decide what to buy and which vendor (seller) to buy it from. Three types of buying situations are possible. 1. Straight Rebuy: In a straight rebuy situation, the buyer knows exactly what to buy and who to buy it from. For example, when Syracuse University needs more toner cartridges for its printers, it will automatically order new cartridges from HP. This ordering is typically done by clerical staff rather than a high level decision maker. Seller’s Perspective: In a straight rebuy situation, fast and reliable delivery is very important. The products involved are often low-margin items, and the ability to distribute large volumes of material efficiently is required for success. 2. New Task: In a new task situation, the customer is purchasing a new product for the first time. For example, a company may be setting up a new athletic facility for its employees. This is typically a lengthy process where committees are formed to decide exactly what the product should be, and multiple vendors are asked to submit bids. There is typically lengthy negotiations before the vendor is selected. Seller’s Perspective: A “new task” situation provides the greatest opportunity to a new vendor to enter the market. As the prospective customer requires a lot of information, the selling team typically includes experts in different areas (product, finance). As negotiations are involved, high level executives from the selling organization often participate in the selling process. The process involves significant risk of failure from the vendor’s perspective. However, 32

if the vendor succeeds and the customer is satisfied with the product, long term relationships are possible leading to modified rebuy in the future. 3. Modified Rebuy: Here, the customer requires small variations from an existing product. For example, the company may like to open an athletic facility at a second office. In these cases, the customer typically has a lot of knowledge about the product, and also has preferred vendors. Seller’s Perspective: Modified rebuy situations typically represent the greatest profit opportunity for the preferred vendor. First, as the customer is already knowledgeable about the product and you as vendor, there is less need to engage experts to educate the customer. Also, the products involved are typically not low-margin items as in straight rebuy situations. 2.6.6 Reason Why Leasing is Common in Business Markets Compared to the consumer market, leasing is more common in business markets because it offers benefits to both the customer and the company leasing the product out. Customer’s Perspective: 1. Tax advantage: Rental payments are totally tax deductible and usually are larger than corresponding depreciation charges on owned products. 2. Convenience: Leased products are often serviced by the party leasing the product out (that is, “lessors”). 3. Less commitment: This has two dimensions: (a) Finance: Your capital is not tied up in an expensive purchase. (b) Technology: You are not tied to a technology that may become obsolete. This is particularly relevant in rapidly evolving technologies. 4. Your demand may be seasonal. For example, a builder may need earth moving equipment only for short stretches of time. Perspective of firm leasing out Product: 1. Total net income is often higher if you lease out rather than sell. This happens because you earn money by servicing the leased out product. 2. You can expand your market by including customers who cannot buy the equipment outright. 3. You can induce trial of a new product by reducing customer commitment. 2.6.7 Organizational Decision Styles Organizations can be placed in four main categories based on their decision styles: 1. Bureaucratic Decision Style. Government agencies usually follow this decision style. Sometimes, institutions like health care organizations or universities also follow this pattern of behavior. These organizations are regulated tightly, and the focus is on making sure all the regulations have been met. 33

Major needs of Customer: The product must meet exact specifications. Proper paper work is required. Low price is also required. (Typically, if the order is large, the sellers have to make bids, and the lowest bid is accepted). Seller’s Perspective: Know when needs arise, the exact specification of the product, and what paper work is involved. Usually all this can be done by the clerical staff. Also, need engineers and accountants to estimate costs are prepare bids. There is not much need for personal selling. 2. Small Companies. For example, a Mom and Pop family diner needs an air conditioner. Characteristics: Small order size. Repeat purchases rarely made. The product is needed immediately. There are usually one or two decision makers who do not have technical knowledge of the product. Major needs of Customer: Fast delivery. Guarantee that the product will work. Service after the sale to make sure the product works properly. Seller’s Perspective: Reduce the customers’s risk. One method used by selling firms in situations like this is to use a regional salesperson or a sales representative to deal with the small customers. It is not necessary for the salesperson (or sales representative) to have a high degree of technical knowledge about the product category as long as she knows how this particular equipment works. 3. Large companies with functional divisions Example: Citicorp purchasing computer work stations for its New York offices. Characteristics: Large orders. Long period of planning and comparing alternatives. Formal buying center consisting of representatives from different divisions (technical, finance etc.). Important: Long term relationships and repeat purchases are possible. Major needs: Quality of product. Satisfaction of clearly defined needs. Depending on the product category, ease of use and scope to expand and upgrade (e.g., computer network) may be important. Seller’s Perspective: Need to have a sales team consisting of technical experts to explain the product’s features to the buyer’s technical people, financial experts to negotiate financial arrangements, etc. The negotiations are likely to be lengthy, and typically the sales team will need to make several presentations. A selling firm often uses a national accounts executive to coordinate the selling activities. 4. Political Decision Style: These involve organizations where participants with political power make purchase decisions. As political power comes from forming coalitions of people with similar interests, purchase decisions also require the support of a group of influential people in the organization. For example, the purchase of a new weapons system often involves political decision style. Seller’s Perspective: You need supporters within the organization to champion your product. Retired government officials and politicians are often hired as lobbyists to build such support. 34

3

Selected Topics on Market Segmentation and Targeting

3.1

Examples of Market Segmentation in Practice: Procter and Gamble

(A) P&G Laundry and Fabric Care: • Bounce: Bounce Sheets can breathe a fresh scent into almost anything, from your favorite sweatshirt to Dad’s old recliner. Bounce Sheets come in four great scents, and an unscented variety. • Cheer: Cheer helps protect against fading, color transfer and fabric wear in powder or liquid, with or without bleach. • Downy: Downy offers a line of advanced fabric care conditioning products that keep clothes feeling soft and smelling fresh. • Dreft: Dreft is a specially formulated detergent that rinses out thoroughly, leaving clothes soft next to a baby’s skin. Dreft has been the No. 1 choice of pediatricians for years. • Era: Era is a powerful laundry detergent that is tough on stains. • Febreze: Febreze is a fabric refresher and air freshener that utilizes a unique odoreliminating technology to clean away odors from household fabrics and the air. • Gain: Gain laundry detergent and fabric softener provide excellent cleaning power and a smell that says clean. • Ivory: Mild cleansing benefits for a gentle, pure and simple clean. • Tide: Fabric cleaning and care at its best (B) P&G Body Wash and Soap (Personal Cleansing Products) • Camay: A moisturizing bar soap enriched with perfumes of French inspiration that leaves your skin feeling fresh, soft and sensual. • Ivory: Mild cleansing benefits for a gentle, pure and simple clean. • Noxzema: Facial cleansers (creams, astringent and pads) deep clean for healthy, naturallooking skin. • Olay: Olay offers an array of skin-care and personal cleansing products that provide multiple benefits designed for women of all ages. 35

• Old Spice: At Old Spice, we’re dedicated to staying up-to-date on emerging technologies that protect you from odor and wetness, and keep you smelling great. With our new scents and constantly expanding line of products, it’s no surprise that Old Spice is a top choice among guys. • Safeguard: Safeguard is the No. 1 antibacterial soap worldwide; it is the only bar soap registered with the FDA. Safeguard is designed to provide excellent germ protection for the whole family. • Zest: Zest is refreshingly different from ordinary soap. It rejuvenates you with a combination of great, refreshing scent and clean-rinsing lather that won’t dry out your skin like soap. Note: You should note two facts about the marketing strategy used by P&G: 1. In a given product category, each P&G product offers a unique benefit. This is the strategy generally used by P&G and is called benefit segmentation. 2. P&G often offers more than one product to the same market segment. For example, in the laundry detergent market, Ivory and Dreft are both meant for cleaning baby clothes, but offer slightly different benefits (Ivory: pure, Dreft: soft).

3.2

Meaning of Market Segmentation

Context: Do the customers all respond the same way to your marketing mix, or are there differences in their response to one or more of the elements of the marketing mix? Examples: Do they have different product needs? Do they respond differently to price changes? Do they patronize different types of stores? Do some customers know a lot about the product category, and others know very little? Are they likely to consume the different amounts of the product? If any of the above holds, market segmentation may be helpful. Formal Definition of Market Segmentation: Market segmentation is the process of dividing a market into meaningful, relatively similar, and identifiable segments or groups based on differences in their response to the marketing mix. Once the segments are identified, the management selects one or more segments as the organization’s target market. Examples of Market Segments: 1. Different Product Needs: Different customers may have different “Ideal Products” or “Ideal Points.” 36

1.(a) Cars: You may describe cars in terms of the two dimensions: fuel economy and ruggedness. For example, Honda Civic hybrid and Toyota Prius rate high on fuel economy but low on ruggedness. In contrast, Hummer and Land Rover rate low on fuel economy and high on ruggedness. As different consumers assign different importance weights to these two attributes, the ideal car for some consumers will be close to a Prius, and the ideal car for some others will be close to a Hummer. 1.(b) Grocery Stores: Grocery stores can be described in terms of two dimensions: price, and width of assortment. For example, Wegmans has moderately high price and wide assortment. On the other hand, BJ’s Wholesale Club has low price and low assortment. Again, some consumers may find Wegmans close to ideal while some others may find BJ’s close to ideal. 1.(c) Location of convenience stores: Generally speaking, a convenience store that is close to where you live or work is preferred to one far away. 2. Different response to price: Are there segments that are more price sensitive than other segments? Some marketing actions: “Special Prices” for students. Coupons and “Sale” prices to attract price conscious consumers. 3. Different response to distribution: A person who enjoys shopping around and visiting malls usually prefer to buy clothes and linen from department stores like JC Penney. On the other hand, someone who either does not enjoy shopping around or does not have time to spare often buy the same items from catalog merchants like LL Bean or Fingerhut. 4. Different response to promotion: Some consumers get their information from the internet while others like to talk to a salesperson.

3.3

Evolution of the Concept of Market Segmentation

The concept of market segmentation has been around for a long time, but it became prominent relatively recently. The evolution occurred in the following three stages: 1. Mass Marketing: Mass produce, mass distribute and mass promote one product to all buyers. Logic: Low cost −→ low price −→ high demand −→ economy of scale Example: Ford (standard black car for $500). 2. Product-variety Marketing: Offer multiple products which may differ on almost any conceivable dimension. For example, Baskin Robbins sells 31 flavors of ice cream. Objective: Offer variety to consumers. Hopefully, there will be customers for each variety offered. Problem: The market may not be interested in all the variations offered. (Would you like to buy roast-beef flavored chewing gum?) Of course, in the long run, it is possible to eliminate the varieties consumers do not want. Still, the process is wasteful. Also, since you are not starting with the customer, you may overlook opportunities. 37

3. Target Marketing: Utilizes segmentation. Steps involved in target marketing: (a) Distinguish between market segments: Market Segmentation. (b) Evaluate each segment and select one or more segments as your target(s): Market Targeting (c) Formulate a competitive positioning for the product and develop a detailed marketing mix: Market Positioning.

3.4

Possible Benefits of Market Segmentation

1. Satisfy Customers: You can meet the target customer’s needs more accurately. This can increase demand for the product in the selected segment. 2. Reduce competitive Pressure: This happens in two ways. (i) Being a smaller market, a segment is less attractive for a competitor to enter. (ii) Customers are satisfied with your product and are less likely to switch. 3. Market segmentation may lead to a more efficient use of your marketing mix. For example, if you are a breakfast cereals manufacturer and have selected children as your target segment, you may advertise during cartoon shows instead of prime time news programs.

3.5

Criteria for Successful Segmentation

Segmentation is only effective if market segments rate favorably on five criteria: 1. Differentiable: Segmentation can be effective only when different parts of the market actually respond differently to your marketing mix. (For example, do some people prefer sweeter soft drinks compared to others?) If there is no such difference, segmentation is not meaningful. 2. Substantial: It is meaningful to develop a marketing mix for a segment only if the segment has large enough purchasing power. This requires commonality, that is, presence of customers who respond similarly to your marketing mix. You can then aggregate the customers into relatively homogeneous subgroups, that is, form segments. Objective: Each segment should be large enough to give you economy of scale. 3. Identifiable and Measurable: Market segments should be measurable in terms of size and purchasing power. It is also desirable to identify the members of the different segments, if possible. 4. Accessible: The segments can be effectively reached and served. 5. Actionable: You, the marketer, must be capable of developing a marketing mix for the targeted segment and reach the segment with the customized mix. 38

Basically, you, the marketer, have to decide if after you segmented a market, can you take marketing actions based on it? For example, can you offer a product which a given segment may want? Can you distribute to this segment? Can you promote your product to this segment through personal selling or advertising? The answer to these questions depend on your strengths and weaknesses, your competitive position, and the method used to segment the market. Also, the number of segments targeted should match your marketing capabilities.

3.6

Bases for Segmenting Consumer Markets

Some bases for segmenting the consumer market are discussed next. In practice, marketers often use combinations of these bases. For example, Geodemographic Segmentation combines geographic, demographic, and lifestyle segmentation. 1. Geographic Segmentation: Consumers in different geographic locations may respond differently to your marketing mix. For example, they may have different product needs, or different levels of price sensitivity. Examples: (i) People in Texas have less need for snow-tires or home insulation than people in upstate New York. (ii) Students in Syracuse have less need for Michigan Wolverine jackets than students in Ann Arbor, Michigan. (iii) Maxwell House Coffee is distributed in different flavors in different regions (stronger in the West). (iv) The grocery store Wegmans offers different assortments of products in different areas. Assessment of Geographic Segmentation: It is easy to measure the size of the population in each geographic segment using Census data. It is also easy to target specific segments with your promotional activities. However, you need to check if consumers in the different geographic areas really differ in their response to your marketing mix. 2. Demographic Segmentation. Examples: Age, Income, Gender, Education, Occupation. The marketer may use multiple variables. For example, the segment of interest may be males between 25 and 40, with college degrees. Assessment of Demographic Segmentation: It is easy to measure the size of the population in each demographic segment using Census data. It is also possible to target specific segments with your promotional activities as different demographic groups often have somewhat different media habits. However targeting a specific demographic segment with promotion is generally more difficult than it is for geographic segmentation. Once again, segmentation will be successful only if the segments respond differently to your marketing mix. 3. Psychographic and life style Segmentation. For example, is a person authority seeking, or is he independent? If a person is authority seeking, you may have to distribute medicine to him through drug stores, and promote your product to doctors and pharmacists. If (s)he is independent, you can distribute the product through department stores, and use television advertisements. Assessment of Psychographic Segmentation: Unlike geographic and demographic seg39

mentation, you cannot measure the size of psychographic segments using Census data. You have to either measure them using survey research, or purchase data from companies like SRI Consulting Business Intelligence (see the discussion of VALS in Section 3.7 of the Reader). 4. Benefit Segmentation: Classify customers based on benefits they are seeking. A segment is a group of customers seeking similar benefits. For example, you may customers with similar attribute importance weights in the same segment. 5. Usage Rate Segmentation: This is discussed quite adequately in the text. The only additional thing you need to note is that heavy users need to be treated differently from the light users because heavy users are often more price sensitive than light users. That can happen due to many reasons. For example, if you are a heavy user, you may pay more attention to your expenses. It can also happen because the market for heavy users is more competitive as it is targeted by more marketers. In recognition of the higher level of price sensitivity, marketers often offer lower prices (through quantity discounts) or incentives (for example, frequent flyer programs) to heavy users.

3.7

Example of Psychographic Segmentation: VALS

A company called the SRI Consulting Business Intelligence, a spin-off of Stanford Research Institute (SRI), conducts an ongoing survey called the “Value and Life Styles (VALS)” survey and segments the US market based on the results. There are different slight variations of the segmentation scheme: I discuss one below. One version of the VALS segmentation scheme: In this version, a person is rated on two dimensions: (1) Orientation: • Principle Oriented: They believe in principles and some form of religion and/or ethical code, and have a strong sense of community responsibility. They generally do not believe in conspicuous consumption. • Status Oriented: They like to achieve status and to demonstrate that status to others. Visible symbols of success are important to these people. • Action Oriented (2) Resources available to the person (money, time, education, etc.): Low to high. Based on the ratings, a person can belong one of eight segments: 1. Innovator (also called Actualizer): Innovators have enormous resources, and are often not targets of mainstream marketers as they get custom made products. 2. Survivor (Struggler): Survivors have very little resources and are again not targets of mainstream marketers. 40

3. Thinker: Principle oriented with high resources. (This group was called the “fulfilled” in earlier versions of VALS.) They engage in philanthropic activities, donate to charities and organizations, and are targets for fund-raising activities.

4. Believer: Principle oriented with low resources. They volunteer for church, community, or political activities.

5. Achiever: Status oriented with high resource. They purchase more expensive symbols of status like expensive homes, vacations, jewelry, and high-end automobiles. Engage in conspicuous consumption.

6. Striver: Status oriented, with low resource. They purchase less expensive symbols of status such as expensive clothes.

7. Experiencer: Action oriented with high resource. They participate in activities such as scuba diving and mountain climbing, purchase performance sports cars, high end music systems and cameras.

8. Maker: Action oriented with low resource. They remodel their own homes, make their own furniture, and targets for do-it-yourself stores like Home Depot in the USA.

A marketer can purchase information such as the size and media habits of a particular segment from SRI Consulting Business Intelligence.

Note:

1. The VALS segmentation scheme satisfies the condition of identifiability and measurability because you can obtain such information from SRI Consulting.

2. You can visit the website sric-bi.com, fill up their survey, and know what is your own type. 41

¡@ ¡ @ ¡

@

¡

High Resource Principle Oriented

¡ Innovator @@ @ ¡ ¡ @ ¡ @ ¡ @ @¡

Status

Action

Oriented

Oriented

Thinker

Achiever

Experiencer

Believer

Striver

Maker

¡

Low Resource

3.8

¡

¡@ ¡ @

@

@ ¡ Survivor @ ¡ @ ¡ @ ¡ @ ¡ @ ¡ @

Strategies for Selecting Target Markets

1. Undifferentiated Marketing: Offer a single marketing mix to all customers. This is sometimes called “Mass Marketing.” Process: First analyze the market, then design a product and a marketing program to appeal to the broadest number of buyers. Idea: Mass market to the largest possible segment. Example: Hershey used to follow the strategy: “One chocolate bar for every person.” Problem: Hard to protect from competition. 2. Multi-Segment Targeting (Differentiated Marketing): Operate in several segments and design a separate marketing mix for each. 42

Possible Advantages of Multi-Segment Targeting: • Consumers will identify company with the product category, e.g., Hewlett Packard with laser printers and Canon with cameras. As a result, a consumer will automatically think of the company when a need for the product arises. • By customizing your marketing mix to a particular segment, you have more satisfied consumers. This leads to a greater chance of repeat purchase and more brand loyalty. Hence you have greater sales volume and reduced competitive pressure. Possible Disadvantages of Multi-Segment Targeting: • Multi-segment targeting is more costly than undifferentiated targeting for various reasons. For example, you may have to: (1) Produce different versions of the product. (2) Develop different promotional campaigns. (3) Distribute the product through different types of channels. • Cannibalization: Your own products may compete with one another for market share. Important Note: Both sales revenue and costs go up when differentiated marketing is used. Examples of Multi-Segment Targeting: • Kellogg: Breakfast cereals • P&G: Different categories of consumer packaged goods, such as laundry detergent • GM: Automobiles 3. Concentrated Targeting (Single-Segment Concentration): Operate in only one segment. This is sometimes called “niche” marketing. This strategy is meaningful when company resources are limited. Note that resources need not always be financial or technological. For example, the resource of a company may be its “image.” Levi Strauss has the “casual” image which it has used successfully to introduce its “Docker” products. That very same image would work against Levi Strauss if it tried to launch an expensive line of formal clothing under the Levi name. (This actually happened when Levi tried to introduce a “Tailored Classics” line in the early 1980’s and failed.) Objective of Concentrated Targeting: Capture a large share of a small market. Example: Porsche (High priced sports cars). 43

3.9

Product Positioning Strategies

Position is “the place a product, brand, or group of products occupies in consumers’ minds relative to competing offerings.” A product’s position is the consumer’s perception of a product’s attributes, use, quality, advantages and disadvantages. More simply, a product’s position refers to what kind of “image” the relevant customers have about the product. A marketer has to actively try to create a position for her product in the mind of the consumer. This is essential whenever a new product is introduced to the market. To position a product, the marketer has to provide the customer a basis for judging the product. This generally requires communication with the customer. Consequently, promotion is very important for a successful positioning activity. Examples of positioning: (a) Cars: BMW has an image of a high priced car with superb engineering. Honda has a image of a moderately priced car with a “good value for the money.” (b) Universities: Harvard University, University of California at Berkeley, and Princeton University are all known as great academic schools. We now describe positioning bases, i.e., the different ways a product may be positioned. Positioning Bases: 1. Position by attribute-benefit: Usually, this is used if a product has a very high rating on an attribute, or offers a unique benefit. Examples: Dr.Pepper has a unique taste. Tums contains calcium. Tylenol does not have any major side effects. 2. Position by price/quality: (a) Quality of the material used to make a product is sometimes used to position a product. Examples: Furniture made of solid oak (versus furniture made of plywood), gold watch. (b) When consumers have very few clear features to compare products on, but have an overall sense of quality, they use price as a measure of quality. Thus, a marketer may use a high price to create a perception of quality. Examples: Fashion by Bloomingdale’s, Swiss watches, French wine, Gucci shoes, perfume. 3. Positioning by use/application: Adidas, Nike, and Reebok are positioned as athletic shoes. 4. Positioning by product user: Create the correct image for a particular market segment. Example: MTV is positioned as a teenager’s channel. Virginia Slims is aimed at women. Example of repositioning: Marlboro for men (Marlboro was originally marketed as a women’s cigarette). 5. Positioning by product class: This can be done in one of four ways. (a) A product can be positioned as similar to a product class, e.g., many Margarine brands were positioned as similar to butter. 44

This strategy, which is a “point of parity (POP)” strategy, is commonly used by companies with efficient production or distribution capabilities who can offer a product at a lower price than its competitors. The focus of the positioning is to offer the same benefits at a lower price. (b) A product can be positioned as being different from a product class: Aleve and Claritin are both positioned as products that work much longer than their competitors. Lexus positions its car based on the unique ability to parallel park itself. This strategy, which is a “point of difference (POD)” strategy, is adopted by companies with cutting edge research and development capability. The focus of the positioning is the unique benefit offered by the product, and a premium price is charged. (c) A product can be positioned as a supplement to a product class: e.g., VCR s were positioned to be supplements to televisions. (d) A product can be positioned as a replacement for a product class: e.g., Nutra-Sweet (as a replacement for sugar). 6. Positioning by competition: This makes sense when you have a major, clearly defined competitor, and your strategy depends on how consumers perceive your product relative to the competition. (a) A product can be positioned as being similar to a competitor. These are generally “me-too products” that do not enjoy a superior image in the mind of the consumer. Thus, the advertising campaign usually tries the convince the consumer that she can get the same product benefits at a lower price. This strategy is routinely used by less expensive automobile manufacturers such as Hyundai. (b) A product can be positioned as being different from a given competitor. For example, two politicians competing for the same votes generally try to position themselves as different from each other. Wendy’s stresses that it uses real chicken meat unlike its competitors. (c) A combination of these two strategies was used by Subaru to sell the “Subaru Legacy”: integrated body design like Volvo, and a better braking system. 7. Positioning by combination of ways: use two or more of the first six methods together. Repositioning: Sometimes, a marketer tries to create a new, more favorable position for an existing product. That is called repositioning. There are many examples of successful repositioning. The College of New Jersey was repositioned as Princeton University. Marlboro, originally a women’s cigarette, was repositioned for men. Rogaine was repositioned as a medicine for hair regeneration rather than blood pressure reduction. Note: Positioning by product class or a specific competitor: If the new product does not have unique features and/or the incumbent is considered superior, the new product usually adopts the me-too strategy. Usually that involves claiming that the new product is just as good as the old product, and that it costs less. For example, 45

Folger positioned itself as similar to expensive coffees purchased from specialty stores, and the brand Natra-Sweet positioned itself as similar to “Equal.” If the new product has unique features, and at least a segment of customers considers it superior because of that, it can stress the difference and charge a higher price. Examples Sanka stressed that it was “decaffeinated” and charged a higher price than regular coffee. Similarly, Nutra Sweet is priced higher than sugar.

46

3.10

Example of Segmentation using student data

The following mini-survey was completed by a sample of 148 undergraduate students at Syracuse University. (These questions are excerpted from the survey provided in Section 11.7.)

1. Gender

Male

Female

2. Are you a member of fraternity or sorority? Yes

No

3. How interested are you in the following activities in your spare time? (Please circle) Not interested at all

Very interested

(a) Exercise

1

2

3

4

5

6

7

(b) Participate in sports

1

2

3

4

5

6

7

(c) Shop for clothes

1

2

3

4

5

6

7

(d) Go to bars

1

2

3

4

5

6

7

(e) Go to malls

1

2

3

4

5

6

7

(f) Watch movie

1 2

3

4

5

6

7

(g) Do volunteer work

1

2

3

4

5

6

7

(h) Study/read

1

2

3

4

5

6

7

(i) Listen to music

1

2

3

4

5

6

7

(j) Spend time with friends

1

2

3

4

5

6

7

(k) Watch sports on TV

1

2

3

4

5

6

7

(l) Watch sports at the Dome

1

2

3

4

5

6

7

47

4. If a combined lacrosse/football/basketball season ticket were available, how likely would you be to purchase it at the following prices? (Please circle) Very unlikely

Very likely

(a) $50

1 2 3 4

(b) $75

1 2

(c) $100

1

(d) $125

1 2 3 4

(e) $150

1

2 3 4 5

6

7

(f) $175

1

2 3 4 5

6

7

(g) $200

1 2 3

3

5 6

7

4 5 6

7

2 3 4 5

4

6

7

5 6

7

5 6 7

**Thanks for Your Participation**

Psychographic Segmentation: The items in question 3 represent lifestyle variables (activities, interest, opinion) of the students. I now use these to do psychographic segmentation using Minitab. Step 1: First, I use a method called factor analysis to find the key underlying dimensions of the items 3(a)-3(l). Four dimensions (factors) emerge. The correlations of these four factors (Factor1-Factor4) with the original variables (X3a − X3l ) are given below. Variable X3a X3b X3c X3d X3e X3f X3g X3h X3i X3j X3k X3l Variance

Factor1 0.681 0.869 −0.223 0.130 −0.167 0.172 0.016 0.038 0.073 −0.055 0.725 0.754 2.4469

Factor2 Factor3 Factor4 −0.059 −0.103 0.328 0.152 −0.070 −0.005 −0.811 0.178 0.075 −0.516 −0.519 0.088 −0.801 0.144 −0.027 −0.547 −0.060 −0.452 −0.121 0.811 −0.002 −0.135 0.691 0.254 −0.092 −0.150 −0.760 −0.565 0.006 −0.384 0.285 −0.057 −0.369 −0.008 0.248 −0.247 2.3339 1.5643 1.3141 48

Each factor has mean zero and the variance given in the last line of the table. Together, they explain 63.8% of the variance of the 12 original variables. To interpret the factors, only look at correlations above 0.5 in magnitude. That way: Factor 1: Interest in watching and participating in sports. Factor 2: Aversion to shopping, bars, and malls. Factor 3: Interest in studying and volunteer work, together with aversion to bars. Factor 4: Aversion to music. Step 2: I now use a method called K-Means clustering to create groups of respondents with similar factor scores. There are four clusters. The average factor score of each cluster is given below. The last row of the table gives the number of respondents in each cluster. Factor Factor1 Factor2 Factor3 Factor4 Size

Cluster1 0.8194 −0.9285 −0.3284 −0.0721 39 (26.35%)

Cluster2 −1.1010 −0.1413 −0.0403 −1.0335 29 (19.59%)

Cluster3 −0.0304 0.0869 1.0193 0.2517 37 (25%)

Cluster4 0.0255 0.8627 −0.5520 0.5458 43 (29.05%)

The four cluster are four psychographic segments with the following characteristics: (1) Cluster 1: Interested in sports, and not interested in shopping for clothes, visiting bars and malls, studies, or volunteer work. (2) Cluster 2: Not interested in sports, shopping, bars, or malls. Interested in music. (3) Cluster 3: Interested in studies and volunteer work. (4) Cluster 4: Averse to shopping, bars, and malls. Not interested in studies, volunteer work, or music. Once we have the segments, we can try to find who belongs to each cluster. For example, we created the following cross tabulation of gender with cluster: Gender 1 Women 15 (20.00%) Men 24 (32.88%)

Cluster 2 3 22 22 (29.33%) (29.33%) 7 15 (9.59%) (20.55%)

4 16 (21.33%) 27 (36.99%)

We can also check if some behavior or purchase intentions differ from cluster to cluster. Consider question 4(e), which is likelihood of purchasing the combined ticket for $150. Let Interest = 1 if the respondent circled 5-7, and 0 if the respondent circled 1-4. The crosstabulation of Interest with Cluster is given below: 49

Cluster 1 2 3 4

Interest 0 1 27 12 (69.23%) (30.77%) 19 10 (65.52%) (34.48%) 16 20 (44.44%) (55.56%) 15 28 (34.88%) (65.12%)

Interesting, cluster 4, which is not interested in shopping, bars, malls, or participating in sports, are most likely to buy the ticket for $150. (There is one missing case in Cluster 3.) Demographic Segmentation: We can also try to relate demographics to usage or intentions. For example, we got the following cross-tabulation of Gender and Interest (1 if X4e ≥ 5, 0 if X4e ≤ 4): Interest Gender 0 1 Women 42 33 (56.00%) (44.00%) Men 35 37 (48.61%) (51.39%) Apparently, men are slightly more interested in the ticket than women. (There is one missing case among men.)

50

4 4.1

Selected Topics on Marketing Research Introduction

4.1.1 Definition of Marketing Research: Marketing research is the function that links the consumer, customer, and public to the marketer through information–information used to identify and define marketing opportunities and problems; generate, refine, and evaluate marketing actions; monitor marketing performance; and improve understanding of marketing as a process. Marketing research specifies the information required to address these issues, designs the method for collecting information, manages and implements the data collection process, analyzes the results, and communicates the findings and their implications. (American Marketing Association, October, 2007). More simply, marketing research is the systematic gathering, recording, and analysis of data with respect to a particular market. Basically, marketing research involves collecting data about the marketing environment so that you can make better decisions about your marketing mix. Information about the macro-environment generally comes from secondary data that has already been collected by someone else. (This is so because many parties are interested in these information, and organizations like the Government or large marketing research companies supply these data). On the other hand, you usually have to collect information about your micro-environment first hand; this is called primary data collection. 4.1.2 Steps in the Marketing Research Process: 1. Identify and formulate problem/opportunity: (a) Problem recognition (b) Exploratory research (c) Problem formulation: develop specific research questions; identify source of your data. 2. Plan research design: (a) Decide on the nature of the study (survey? experiment? focus group?) (b) Select nature of data collection (ask questions? observe behavior without asking? combination of the two?) 3. Specify sampling procedure 4. Collect data 5. Analyze data 6. Write report 7. Follow up 4.1.3 Data typically collected from customers The purpose of a marketing research study is usually to understand and predict customer behavior. The data collected can be broadly divided into two categories: 51

• Non-Behavioral data, which are not on customer behavior itself, but which can help understand customer behavior. • Behavioral data: Data on some element of customer behavior. A) Non-behavioral Data: 1. State of being variables: • Demographics: Basic descriptors of the customer, such as gender, age, income, marital status, and geographic location. • Ownership status: Ownership of items that may influence customer behavior. For example, if a student has a car, he may be more likely to visit bars and restaurants away from campus. 2. State of the mind variables: • Does the customer know about your brand? This is called awareness. Awareness can be measured at two levels: – Top of the mind awareness, where no brand name is provided. For example, the question may be: “Name the first brand of laundry detergent you can think of.” Suppose you wish to measure the top of the mind awareness of Cheer. If a respondent names Cheer, she gets 1, and if a respondent does not name Surf Excel, she gets zero. For the aggregate sample, you compute the fraction of the sample that named Cheer. Note that even within this narrow definition, a change in wording may give very different results. For instance, if you asked the respondent to list all the brands she can think of, the top of the mind awareness of any given brand will be higher than is you asked “name the first brand you can think of.” – Aided recall, where the respondent is given a brand name and then asked if she had heard of that brand. It becomes 1/0 variable for a given respondent (1 if the respondent has heard of the brand, 0 if not), and a fraction for the aggregate sample (proportion of the sample that has heard of the product). Generally, top of the mind awareness is used for well-known brands, and aided recall is used for new of unfamiliar brands. • How important are product attributes to a customer? • How does the customer rate your brand on different attributes? • How much does the customer like or dislike your brand? 52

Life style variables (activities, interests, opinions) of the customer may also be included as non-behavioral variables if they have an effect on the customer behavior we are interested in. B) Behavioral Data: 1. Usage behavior (past, present): Does the customer buy in the product category? How much money does he spend? Where and when does he buy the product? Which brand or brands does he buy? How brand-loyal is he? Does the customer buy your brand? 2. Behavioral intentions: Does the customer plan to buy your product in the future? Will he recommend your product to a friend? 3. Media habits Use of These Information The focus of a marketing research study is usually some form of customer behavior, such as how much money one spends on different brands of soft drink. • By aggregating expenditure information from a sample of respondents, the researcher can estimate the demand for the product category. • By aggregating expenditure information from a sample of respondent, the researcher can estimate demands for specific brands. • By combining the above two, the researcher can estimate market shares of brands. • By tracking sales over time, the researcher can determine if the sales are growing or shrinking over time, and whether there is any seasonal variation. • By tracking the purchase behavior of a given person over time, the researcher can determine whether the person tends to buy the same brand again and again, or does (s)he buy whatever is available on sale. Such information can be used to plan production and set an appropriate pricing strategy. Thus, knowing just the usage pattern can be quite useful to a marketer. • By tracking the purchase behavior of a fixed sample (called panel) of persons or households over time, the researcher can construct a brand-switching matrix. The knowledge of other variables measured helps the marketer refine its actions. For example: • A study of media habits helps the marketer determine where and when it should advertise. • By examining the relationship between demographic variables and usage behavior, the marketer can determine exactly who buys its products and which demographic segments represent the greatest untapped potential. 53

• A study of awareness tells the researcher how many people know of the product. • A study of importance weights of attributes and ratings of brands on attributes tells the marketer how his brand is performing on relevant attributes. From such a study, the strengths and weaknesses of the brand can be determined, and directions for product improvement or new product development identified. Note that a piece of information should be collected only if it can be used to improve marketing decision making. Otherwise, it is of no value to the researcher. 4.1.4 Sources of Secondary Data: Secondary data are data that have already been collected, not specifically for your study. These can come from internal and external sources. Internal Secondary Data: These are data already collected by your organization. Examples include data on sales, costs, prices, and sales call reports. External Secondary Data: These can be of three types: 1. Data from library and internet sources. 2. Government sources (e.g., census data). For example, data collected by the US Census of the Population, conducted every ten years, provides city-block level data on number of households, average income, education, ethnic composition, and nature of households. Retailers routinely analyze these data to estimate the demand for different products in a given geographic area. A particularly important Government source is the data obtained by the US Census of Retail Industries, conducted every 5 years. For example, you may visit the site www.census.gov/econ/census02. This gives you, for every NAICS category, the number of establishments, dollar revenue, payroll, and number of employees in the USA, and also region by region. This information is used to make sales-force allocation decisions. 3. Syndicated services. These are large marketing research firms such as Nielsen and Information Resources Incorporated (IRI) that provide data to paid clients. For example, Nielsen publishes the market shares of television programs at different time slots for different demographic groups in the US population.

Exhibit: Examples of Syndicated Services A. Retail Data: Suppose you are a company like Kellogg’s which sells products like breakfast cereals to customers all over the USA and sometimes worldwide. Naturally, it would like to know how much of each product offering it sells each month, and how the sales are affected by marketing actions such as commercials, coupons, new package design, etc. However, these tasks are complicated by the fact that companies like Kellogg typically do not sell the products directly to the end customers. Rather, it sells the products to marketing intermediaries who, in turn, sell them to other resellers or end customers. While Kellogg’s 54

can monitor when the product leaves its warehouses, it does not know exactly when the sales to the end customers occurs or who purchases them. There are several syndicated services which provide such information to companies like Kellogg’s. However, even these services differ in terms of the unit of measurement, that is, who the data are collected from. The term retial data is used when the unit of measurement is a retail store, and data are collected on sales made by the store in a specified time period. We now discuss retail data provided by the Nielsen Company. The services offered by Nielsen have evolved and expanded with time. Instead of discussing the range of those services which are easily available on the internet, I will describe how Nielsen used to collect these data until the 1980’s. The method is very robust, and can be used in both advanced and developing economies. The Nielsen Company maintained a fixed panel 1 of retial stores from the USA and measures the sales of a wide variety of packaged goods including food (e.g., yogurt, coffee, ketchup) and personal care items (e.g., toilet tissue, detergent). A disproportionate stratified sample with 1600 supermarkets, 750 drug stores, and 150 mass merchandisers was used. Store audits were conducted at two month intervals. During an audit, the auditor recorded the inventory of each stock keeping unit (SKU) as well as how many units of the product were shipped in since the last audit. You can think of the data from a given store as a spreadsheet. Each row of the spreadsheet are the data collected on a single audit. Two columns of data come from each SKU: (1) current inventory, and (2) shipments since the previous inventory. Each audit adds a new row to the spreadsheet. For a given SKU, sales in a given period is computed as: Beginning Inventory + Shipments Received in Period − Ending Inventory This is a well known formula in accounting, and a moment’s reflection shows how much easier it is to use this formula instead of tracking day to day sales. By measuring the sales of a given SKU (e.g., 64 oz Hunt’s tomato ketchup) in a specific time period from each store in the sample, Nielsen estimated the total sales of the product in the US market in that time period. Also, as the subscribing company got data about the sales of its competitors’ products, it could determine the market shares of its products and track how these were changing over time. Also, by combining sales data from Nielsen with its own internal secondary data on promotional activities, a company could determine how sales depended on its marketing decision variables. In addition to the above information, Nielsen also collected data on the shelf facing of each SKU, and whether the retail store did any local advertising for the SKU in the time period concerned. From this information, a company could determine how sales depended on shelf facing and local advertising. There is one limitation of the Nielsen Retail Audit method as described above. The quantity measured by the retail audit formula includes actual sales plus items lost or stolen, 1

A fixed panel of stores means a sample of which remains same over time and are measured again and again. Similarly, a fixed panel of households means a panel consisting of the same group of households that are measured repeatedly. Note that maintaining a fixed panel is difficult as some members may drop out over time.

55

which is called shrinkage by the retail industry. Thus, the retail audit formula provides an over-estimate of sales. To correct this bias, we can measure sales by a more costly and more expensive method such as day by day recording of sales, and determine the percent overestimation in the retail audit formula. With this knowledge, it is possible to obtain approximately correct sales from retail audit results. Example:

Suppose you have conducted monthly audits of a local grocery store where

you visit the store the first day of each month and, for each stock keeping unit (SKU) record: (1) present inventory, and (2) shipments received since the last inventory. The records for Kellogg’s Rice Krispies 19 oz packages for January 1, 2007 through June 1, 2007 follow: Date Recorded

Present Inventory

January 1, 2007 February 1, 2007 March 1, 2007 April 1, 2007 May 1, 2007 June 1, 2007

400 600 600 500 300 400

units units units units units units

Shipments Received Since Last Inventory 500 units 400 units 300 units 500 units 400 units 500 units

From these data, we can determine the sales in months January, February, March, April, and May. Consider for example the month of January, that is, the times period between January 1 and February 1. On January 1, the inventory is 400 units. During January, 400 additional units were shipped in (shipments recorded on February 1). Thus, if there were no sales, there would be 400 + 400, that is, 800 units in inventory on February 1. The difference between this number (800) and the actual inventory on February 1 (600) is the number of units sold in January, that is: Sales in January

=

400 + 400 − 600 = 200 units.

We can also get this result by the direct application of the retail audit formula. B. Consumer Purchase data: Consumer data differs from retail data in that the unit of measurement is a household rather than a store. Here, a fixed panel of household is tracked for a long time, and purchases of a large number of packaged goods are recorded. Note that in contrast with retial data, consumer data tracks a given household over time. These can, for example, tell a company precisely who is buying its products, and whether customers are brand loyal. We now provide three examples of consumer panel data, where example B.1 is a traditional method, and examples B.2 and B.3 are new, improved versions of B.1. B.1. National Purchase Diary Panel (NPD): The NPD used a fixed panel of 13000 households, selected to represent the US population on major demographic characteristics. A member household keeps a detailed monthly diary of the purchase in 50 product categories and records brands purchased, quantity purchased, store purchased from, price paid, whether any coupon was used, and intended use of the products. The completed diaries were mailed 56

to NPD at the end of the month. From this information, NPD estimated the sales of different brands in the USA every month. One can easily see that a consumer may fail to record the purchases on a timely basis and thus make errors as they try to complete the surveys from memory. Also, as the task is tedious, omissions may occur. Further, as the consumer has to report prices paid, (s)he may pay more attention to price and thus become more price sensitive over time. The following two methods try to eliminate such errors. B.2. IRI Scanner Panel Data: These data are provided by Information Resources Incorporated, known as IRI. IRI maintains fixed panels of households in several medium sized towns such as Marion, Indiana. In each town, a panel of about 2000 households is maintained. The towns selected are self contained communities where residents purchase all groceries within the community. IRI issues cards to the households and makes arrangements with the grocery stores to accept the cards. Each time a member household makes a grocery purchase, the card is scanned at the check-out counter, and details of the purchase such as brand purchased, price paid, coupon used, etc., are recorded automatically. This avoids errors of omission that can occur in a purchase diary that the customer has to fill. Further, as the respondent is not providing the information actively, her/his purchase behavior would not change with time. B.3. New NPD Method: To compete with IRI, NPD introduced an a variant of its diary panel where a panel member household is given a wand attached to a black-box. The consumer is asked to draw the wand across the bar codes of the packages after each purchase. Details of the purchase are automatically recorded. Note that this method can be used in all types of communities, and is not restricted to small self contained communities like the IRI. As these examples show you, marketing research is a very dynamic field, and you have to upgrade your services constantly if you wish to survive. C. Media habits. We will discuss only one example, the well known Nielsen TV ratings. The Nielsen Company uses a panel of about 1000 US households, selected to represent the US population in terms of ethnicity, income, age distribution, etc. Before 1988, the member household had a device called the Audimeter, attached to the television sets, that automatically recorded when the TV was turned on and what channel was watched. From this information, Nielsen determined and published the market shares of different TV programs at each time slot of the day. The Audimeter could not record who was watching the programs. Thus, since 1988, Nielsen has used a modification of Audimeter called the Peoplemeter. Each viewing member of a panel household is given an ID which (s)he is supposed to enter when watching the TV. That way, Nielsen can report the market shares of programs for different age groups also. Do you see any problems with this approach?

4.1.5 Types of Research Studies: Depending on your needs, you may conduct three types of research studies: exploratory research, descriptive research, and experiment. 57

4.1.5.1 Exploratory Research: This is usually done when you are still formulating your research questions and want to make sure you do miss anything of importance. This requires a flexible research design. Three important methods of exploratory research are as follows: (1) Analysis of Selected Cases: Two groups of cases that represent two extremes of outcome are selected. The groups are then compared attribute by attribute to see where they differ. These are the relevant attributes. For example, in Section 1.5, Sprint and Verizon represent two extremes of outcome: low market share and high market share. Sprint has higher ratings on three attributes: price of handset (b), long distance rate (c), and features included (e). Verizon rates higher on monthly fee (a) and reliability of connection (d). Clearly, (a) and (d) are more important than (b), (c), and (e). (2) Focus Groups: • A group of people are brought together in a moderated session. The people may be present or potential customers, or people who are not customers but can provide information about the decision making process. • A typical session runs for 1.5-2 hours and include 8-12 respondents. Student sessions are usually shorter (about 1 hour) and smaller (about 5-6). • The moderator starts the conversation and gets the group going. She has an outline but not a fixed questionnaire. She initiates a topic, gets the conversation flowing, and probes further if interesting issues come up. She has to avoid conflicts by changing the topic if necessary. She also has to identify people who are not talking and get their views. She does not take notes. • Sessions are usually recorded. Professional focus groups are conducted in special rooms with one-way mirrors. The recorders sit behind the one-way mirror. • Focus groups are suitable for idea generation but not statistical analysis. Uses of focus groups include: – Identify what customers like or dislike about your product or company. – Identify the attributes that matter to customers. – Show customers new product concepts and get their reaction. These are particularly useful in weeding out product ideas that are likely to fail. – Show customers commercials in cartoon form and get their reactions. (3) Study of secondary data. 4.1.5.2 Descriptive research: Here, you have a population of interest, and you want to measure a pre-determined set of variables from this population. Descriptive studies have the following basic features: 58

(1) The problem is well structured and variables of interest are known and defined. (2) Since the variables are pre-determined, a standardized formal questionnaire can be used. Cost/respondent is low, which allows you to collect large, representative samples and perform statistical analysis. Common examples of descriptive studies are: • Survey research (examples: mail survey, telephone survey, online survey, household doorto-door survey, and group setting). • Panel data: A fixed set of respondents are measured repeatedly. Examples: purchase diaries, shopping cards. 4.1.5.3 Experiment: The purpose of an experiment is to determine the effect of a change in your marketing mix on outcome variables of interest. Example: A major printer/copier company wanted to find out the following: • Does adding color to business documents improve information retention? • Does adding color to business documents result in greater recall of information? • Do people respond better and/or faster to business documents with color (invoices, call to action materials, etc.)? Here, the change in the marketing mix is: Adding color to a document You wish to measure changes in the following outcome variables: (1) Information retention (2) Recall of information (3) Speed of Response (4) Rate of Response Other Examples: 1. Determine the effect of a price-cut on units sold. 2. Determine the effect of an advertising campaign on what percentage of the population know about your product (awareness level). 3. Effect of an advertising campaign on the sales of a product. Controlled Experiment: In conducting an experiment, a marketer must try to make sure that the change in outcome (e.g., units sold) is happening due to the change in marketing mix (e.g., price cut), and not some other factor (e.g., change in the weather or drop in interest 59

rate). To eliminate such other factors, marketers usually conduct controlled experiments where two groups are compared: (1) An experimental group where the marketing mix is changed. (b) A control group where the marketing mix is not changed. Example 1: To measure the effect of price cut on units sold, a chain of retail stores may introduce the price cut in some stores (experimental group) and keep price unchanged in some the other stores (control group). By comparing sales in the experimental and control groups, the marketer can determine how much sales changed due to the price cut. In this context, a basic experimental design can be expressed as follows: (R) EG (R) CG

O1 O3

X O2 O4

EG represents the experimental group consisting of a sample of stores where the price was cut, and CG is the control group (the group of stores where price was not changed). X is the exposure to the treatment, that is, the price cut. O1 and O2 are observations (average sales/store) in the experimental group before and after the treatment. O3 and O4 are observations (average sales/store) in the control group before and after the time the treatment is applied to the experimental group. (R) represents the fact that the stores were assigned randomly to EG and CG. The effect of the treatment is measured by (O2 − O1 ) − (O4 − O3 ). Example 2: To find if a respondent pays more quickly if she sees the utility bill in color instead of black and white, a utility company may send the bill in color to one group of clients (experimental group), and in black and white to another group of clients (control group). A basic experimental design can be expressed as follows: (R) EG (R) CG

X

O1 O2

Once again, EG is the group of clients exposed to the treatment (bill in color), and CG is the control group (bill in black and white). (R) represents random assignment to groups. O1 and O2 represent observations (average time to pay bill). The effect of the treatment is measured by (O1 − O2 ).

4.2

Basics of Sampling

Important: Please ignore the discussion of probability samples and non-probability samples in the text (bottom of page 237 to end of page 238). 4.2.1 Basic Idea: In sampling, our objective is to study a specific population. A sample is a subset of the population that we study to determine what the entire population is like. For example, suppose we plan to survey the Syracuse University undergraduate student population (approximately 11,000 students). If we draw the names of 100 students from the 60

student directory and study them, then these 100 students would be our sample. Sampling methods include non-probability and probability sampling techniques, as explained next. 4.2.2 Non-Probability Sampling In non-probability sampling techniques, the probability that a given member of the population will be selected is unknown. I will discuss four specific non-probability sampling techniques: convenience sampling, quota sampling, judgment sampling, and snowball sampling. 4.2.2.1 Convenience Sampling: Here, the criterion of choosing the sample is convenience. For example, I may collect data from this class because you happened to be there. Similarly, we may collect data from shopping malls, friends, and class-rooms simply because they can be accessed easily. However, it is not always a bad technique to use. Consider two examples: (1) I am using your class to estimate what percentage of Syracuse University students own Nike shoes. (2) I am using your class to estimate what percentage of Syracuse University students read the Wall Street Journal at least once a week. Clearly the sample will give me reasonable results in case (1) but an over-estimate in case 2. The reason is as follows: You are convenient for me to access because you are interested in business management. Consequently, you are more likely to read Wall Street Journal than other Syracuse University students. Stated differently, the reason for convenience here is correlated with interest in business and hence reading the Wall Street Journal. There is no such correlation between being a business student and using Nike shoes. Thus, using a convenience sample is reasonable as long as what you are measuring is not correlated with what makes the sample convenient to access. 4.2.2.2 Quota Sampling: Here, the sample is selected to have the same composition as the population in terms of one or more relevant characteristics. For example, to determine what percentage of Syracuse University students read the Wall Street Journal, I could collect a sample that has the same proportions of business students and others as the overall Syracuse University student population. Note that a characteristic is relevant only if it is correlated with what you are measuring. If there was no correlation between reading the Wall Street Journal and being a business student, then it would not matter if you draw your sample from business students or others. 4.2.2.3 Judgment Sampling: Use your judgment to get a representative sample. For example, market researchers often conduct test markets of new products in Syracuse or Columbus, Ohio because they feel these areas behave largely like the overall USA. 4.2.2.4 Snowball Sampling: This method is used to collect data from a small population embedded within a large population, such as people who share a hobby, or belong to the same ethnic group. This method works well when people within the small population know each other. In this method, you first identify a member of the population and collect data from 61

her. Then you ask her to identify other members of that population. Thus, your sample “snowballs.” 4.2.3 Probability Sampling In probability sampling techniques, every member of the population has a known and non-zero probability of being selected. The probability of selection may or may not be equal for all members of the population. A researcher using probability sampling draws the sample from a frame, which lists all members of the population either individually or in groups. We discuss three categories of probability sampling techniques: simple random sampling, stratified sampling, and cluster sampling. A special case of cluster sampling, called systematic sampling, is discussed in detail. 4.2.3.1 Simple Random Sampling: In this method, the researcher obtains a complete list of population members, and draws one member of the sample at a time. In each draw, anybody who has not been selected yet is equally likely to be chosen. The sample chosen this way has two properties: 1. Every member of the population has the same chance of being selected. 2. Every group of population members is as likely to be selected as every other group of the same size. Limitation: The second property of a simple random sample (every combination equally likely) is not always desirable. As a specific example, suppose there are 10,000 grocery stores in a state, out of which 2000 are large stores, and 8000 are small stores. You want to select a sample of 100 stores to estimate the average annual sales of Coca Cola by a store in this state. If you select a simple random sample of 100 stores, this sample can be any combination of 100 stores, including all large stores and all small stores. Thus, your estimate will vary widely depending on which 100 stores are chosen. If you only got a list of the 10,000 stores and do not know which are small and which are large, you cannot do any better than the simple random sample. However, if you have separate lists of small and large stores, stratified sampling discussed next is better than simple random sampling. 4.2.3.2 Stratified Sampling: In this method, the population is first divided into sub-groups called strata or layers. Then, a separate simple random sample is selected from each layer. Formally, we proceed as follows: 1. Divide population into mutually exclusive and collectively exhaustive subgroups (strata). 2. Choose a separate simple random sample from each stratum. 3. Compute a weighted sum of the estimates from the different sub-groups. The estimate from a subgroup gets a weight equal to the proportion of the population that belongs to the subgroup. Example 1: Suppose again there are 10,000 grocery stores in a state, of which 2000 are 62

large stores and 8000 are small stores. Suppose you have drawn two separate simple random samples: (1) 50 large stores. Average annual sales of Coca Cola is $30,000. (2) 50 small stores. Average annual sales of Coca Cola is $4000. The estimate of the average annual sales of Coca Cola for the entire population of stores is:

2000 8000 × 30, 000 + × 4000 = (.2 × 30, 000) + (.8 × 4000) = 9200. 10, 000 10, 000

Example 2: 20% of Syracuse University undergraduates are business majors, and 80% are non-business majors. You have conducted stratified sampling with business majors and nonbusiness majors as the two strata. Results: (1) 100 students with business majors. Of these 100 students, 60 students read Wall Street Journal every week. (2) 100 students with non-business majors. Of these 100 students, 25 students read Wall Street Journal every week. Estimate of the fraction of Syracuse University undergraduate students who read Wall Street Journal every week: (0.2 ×

60 25 ) + (0.8 × ) = (0.2 × 0.6) + (0.8 × 0.25) = 0.32 100 100

Important: Stratified sampling gives a better estimate than simple random sampling only when the different strata are different from one another in terms of the quantity of interest. If all the strata are similar, then it does not matter which one you draw the sample from, and there is no improvement over simple random sampling. Thus, when creating the strata, you should try to make them as different as possible. 4.2.3.3 Cluster Sampling: Cluster sampling involves the following steps: 1. Divide population into mutually exclusive and collectively exhaustive blocks (clusters). 2. Randomly choose one or more clusters. 3. Restrict your attention to the clusters you have selected in step 2. Either include all members of these clusters in your final sample (one-stage cluster sampling), or choose a random sample from these clusters (multi-stage cluster sampling). Important Note: In cluster sampling, the final sample comes only from the cluster or clusters selected in step 2. Thus, this method works best if the clusters are similar to one another and hence similar to the overall population in terms of the quantity of interest. One-Stage Cluster Sampling: In one-stage cluster sampling, after you select the cluster or clusters in step 2, you include every member of the selected cluster(s) in the final sample. 63

For example, suppose you want to study the population of SU students who live in dorms, and have selected the dorms as the cluster. If you randomly choose one dorm and include every student in that dorm in your sample, that would be one-stage cluster sampling. Multi-Stage Cluster Sampling: In multi-stage cluster sampling, after you select the cluster or clusters in step 2, you again randomly choose some members of these clusters. For example, in a common method used by economists, a city is divided into blocks (clusters). Some city blocks are selected randomly. Then, from each city block chosen, a fixed percentage of households are chosen randomly. This is an example of two-stage cluster sampling, and is called city-block sampling. 4.2.3.4 Systematic Sampling: Systematic sampling is an example of one-stage cluster sampling. In this method, you first obtain a complete list of population members. For example, suppose you got a list of the 11,000 undergraduate students at Syracuse University, and want to draw a systematic sample of 100 students from it. Proceed as follows: 11, 000 1. Compute the ratio of population size to sample size: K = = 110. K represents 100 one in how many in the population you are selecting, and is called the skip interval. (More generally, if you are selecting a sample of size n out of a population of size M , the skip interval M is .) n 2. Randomly choose a number from {1, 2, . . . , K} such that each number is equally likely to be chosen. Call this number R. 3. From the list of population members, pick the following as your sample: R, R + K, R + 2K, R + 3K, . . ., R + (n − 1)K. To see that this is one-stage cluster sampling, write the list of 11,000 students as follows: 1

2

3

...

110

111

112

113

...

220

221

222

223

...

330

.. .

.. .

.. .

.. .

.. .

10,891 10,892 10,893 . . . 11,000 Note that each column in the above table is a cluster, and each cluster is equally likely to be selected. If a cluster is selected, then every member of the cluster is included in the final sample, that is, this is one-stage cluster sampling. Note: 1. In systematic sampling, every member of the population is equally likely to be selected. For example, in the example given, every member of the population has the same probability, 64

1/110 of being selected because the cluster she belongs to has the same chance, 1/110, of being selected as any other cluster. However, not all combinations are equally likely. For example, members 1 and 2 in the list will never be selected in the same sample. 2. Systematic sampling gives much better results than simple random sampling if you can create a list where the quantity of interest either goes up or down along the list. For example, suppose instead of 11,000 students, you have a list of 11,000 grocery stores sorted from the smallest store to the largest store by floor size. You will use the sample of 100 stores to estimate average annual sales of Coca Cola. Here, a systematic sample is guaranteed to include stores of all sizes. (In contrast, a simple random sample of 100 stores could potentially consist of all small or all large stores.) Clearly, sales of Coca Cola will be larger for larger stores. Since the systematic sample includes stores of all sizes, it gives a more reliable estimate of average sales of Coca Cola than a simple random sample of the same size.

4.3

Non-Response in Mail Surveys

In one-time mail surveys, the typical response rate is only about 20%. This is a problem if people who respond to the mail survey are systematically different from people who do not respond. For example, suppose you are conducting a mail survey to determine what percentage of Syracuse undergraduates are interested in taking a class to learn the statistical package SAS. Clearly, students who are interested in learning SAS will be much more likely to respond to your survey than students who are not interested in the class. Thus, there are two types of students: the “responding types” who are more likely to be interested in the class, and the “non-responding types” who are less likely to be interested in the class. Note that you are only getting responses from the “responding types.” Clearly, if you use this sample alone, you will overestimate the percentage of Syracuse undergraduates who are interested in the class. In other words, non-response creates a “biased sample” here. To address this problem, researchers use a two-step approach: (1) Improve response rate: Personalized cover letter, stamped self addressed return envelope, multiple mailings of questionnaire, incentives such as donation to charity. (2) Correct for non-response: To do the correction, you have to draw a sample from the non-respondents. This is done by using some other method, typically a telephone survey. Basically, you randomly call about one-third of the original sample and ask them if they have had a chance to return the survey. If they have, thank them and hang up. If not, ask them if they could spare a few minutes answering the survey on the phone. At the end of the process, you have two samples: (1) The original sample obtained by mail. This is a sample of the “responding types.” (b) A second sample drawn from people who did not respond by mail. This is a sample of the “non-responding types.” 65

The results are combined as illustrated below. Example 1: Suppose Syracuse University (SU) is trying to decide whether it should build a parking garage exclusively for students. To do that, it must charge an additional fee to the students. SU therefore wants to determine how much an SU student will be willing to pay in additional fees/semester for this service. Suppose 400 questionnaires were mailed out, and 100 came back. In this sample, students were ready to pay a fee of $200/semester on the average. Clearly, the respondents would be more likely to be interested in this service, and $200 would therefore be an overestimate of how much an SU student would be willing to pay for the service on the average. To correct for non-response, suppose you collected data from 50 of the non-respondents by telephone. On the average, these students are willing to pay $50 a semester for the service. You now have two samples: Sample (1): 100 initial respondents. Average = $200. Sample (2): 50 initial non-respondents. Average = $50. Note that the initial response rate was 100 out of 400, that is, 25%. Therefore, it is reasonable to believe that 25% of the students are willing to pay $200 a semester for the service. Similarly, 75% of the original sample did not return your survey. Thus, it is reasonable to believe that the other 75% of the students are ready to pay only $50 a semester for the service. Combining, on the average an SU student would be ready to pay: 0.25 ∗ 200 + 0.75 ∗ 50 = 87.5, i.e., $87.50 a semester for the service. Example 2: A company is trying to determine what proportion of Syracuse University students are interested in an online movie download service that charges $20/month for unlimited downloads. It mailed surveys to a random sample of 400 students. 80 came back, and 40 out of the 80 were interested. 100 out of 320 students who did not return the mail survey were contacted by telephone, and 10 out of the 100 were interested. You now have two samples: 40 = 0.5. 80 10 Sample (2): 100 initial non-respondents. Sample proportion interested = = 0.10. 100 80 = 0.2. Thus, about 20% of the population are like the people The initial response rate is 400 who responded to your survey, and 80% of the population are like the people who did not respond to the survey. Combining, an unbiased estimate of the proportion of SU students interested in the service is: Sample (1): 80 initial respondents. Sample proportion interested =

(0.2 × 0.5) + (0.8 × 0.1) = 0.18 Generalization: 66

In general, let Pr be the response rate to the original mail survey. Thus, approximately the fraction Pr of the population are “responding types,” and the remaining (1 − Pr ) fraction are of “non-responding types.” In Example 1, Pr = .25, and (1 − Pr ) = .75. The corrected estimate is: Pr ∗ estimate from the initial respondents + (1 − Pr )∗ estimate from the second sample drawn out of non-respondents.

4.4

Example: Focus Group Outline

Instructions to Moderator: A. (5 minutes) Introduce yourself and ask their names. State that the focus group will discuss their lifestyle and interests, but do not say this is about Fossil yet. Ask them to help themselves to snacks and drinks. B.(5 minutes) Personal background: major, residence on campus, home state and hometown, age. C.(5 minutes) How do they spend leisure time at SU: Hang around with friends? Go to malls? Watch TV? Go to bars? D.(5 minutes) Media habits: radio, TV, newspapers, magazines E.(10 minutes) Favorite TV commercials: humorous, slice of life, dramatic Favorite magazine commercials: Invite them to discuss which ones they like and why F.(10 minutes) Designer brands: Do they own any? (First let them come up with their answers. If they do not, then suggest product categories: clothes, watches, pens, etc. You may also suggest brand names such as Tommy Hilfiger, Guess. Do not mention the Fossil name yet.). If yes, then which brands do they own? Which brands do they like best? What are they looking for when they buy designer products? Where do they buy them? Do they always buy the same brands? Reason for purchase: own use, gifts. G.(10 minutes) State that the focus group is about Fossil products: Do they know about Fossil products? Do they own any? Do they like them? Have they seen any Fossil commercials? Where? Do they like them? Which products do they think meet the Fossil brand image? Which other brand is most like the Fossil brand? H.(5 minutes) Ask for suggestions about: (1) What can Fossil do to make its products more attractive? (2) How can Fossil reach them (that is, people like the focus group participants)?

67

5

Conjoint Analysis

5.1

Idea

Like the expectancy value model, conjoint analysis assumes that the evaluation of a product alternative depends on the attribute profile of the alternative, (X1 , . . . , Xn ). The evaluation score, or utility, is denoted by U (X1 , . . . , Xn ). We discuss the simplest form of this utility function, called the additive part-worth utility model: (5.1) U (X1 , X2 , . . . , Xn )

=

U0 + U1 (X1 ) + U2 (X2 ) + . . . + Un (Xn ),

where U0 is a constant term, and U1 , U2 , etc., are functions of the different attributes. This model assumes that the marginal effect of a change in an attribute (say X1 ) on utility does not depend on the levels of the other attributes (i.e., X2 , X3 , etc.). Still, the conjoint model is more general than the expectancy value model. Conjoint analysis examines products that can be expressed in terms of attributes that are easy to describe to respondents. In general, the attributes may be qualitative or quantitative. For simplicity, we only discuss quantitative attributes. The analysis is done at the level of an individual respondent who is asked to rate several hypothetical products on a given scale. From the evaluation scores, the researcher can determine how the respondent’s evaluation of the product depends on attribute levels. This information is used to guide new product development and to assess the potential market shares of new product concepts. Meaning of Part-Worths: The functions U1 (X1 ), U2 (X2 ), etc., are called the part-worth scores of the different attributes. We consider a fixed range of each attribute, and use a simple convention: At the lowest point of the range of an attribute, the part-worth score of the attribute is zero. With this convention: • U0 is the evaluation score of the product for which each attribute is at the lowest point of its range. • Ui (Xi ) is zero if Xi is at the lowest level of its range. Otherwise, Ui (Xi ) is the marginal change in evaluation score if attribute Xi is higher than its lowest level. Example: Consider a product with 2 attributes: a laptop computer with attributes weight (X1 ) and hard drive size (X2 ). X1 can rage from 4 lb to 10 lb, and X2 can range from 80 GB to 160 GB. The laptop is evaluated on a 0-100 (very poor to excellent) scale. In this case, the additive part-worth model can be expressed as follows: (5.2) U (X1 , X2 ) = U0 + U1 (X1 ) + U2 (X2 ), where U (X1 , X2 ) is the evaluation score of a laptop with attribute levels (X1 , X2 ). By our convention, we assume that: 68

• U1 (4) = 0 • U2 (80) = 0 Under these assumptions: U (4, 80)

=

U0 + U1 (4) + U2 (80)

=

U0 + 0 + 0 = U0

Therefore: • U0 is the evaluation score of the “basic” laptop computer with each attribute at the lowest levels in the ranges considered, that is, X1 = 4 and X2 = 80. • U1 (X1 ) is the marginal adjustment to evaluation score depending on the level of weight (X1 ). If X1 = 4, then no adjustment is needed. • U2 (X2 ) is the marginal adjustment to evaluation score depending on the level of hard drive size (X2 ). If X2 = 80, then no adjustment is needed. Estimation: To estimate the conjoint model, the researcher creates hypothetical products, and asks the respondent to rate each hypothetical product on a given scale (e.g., a 0-100 rating scale). To develop the hypothetical products, a fixed number of levels of each attribute is used. For a given attribute, the lowest and the highest point of its range are always used. Once the respondent rates the hypothetical products, the ratings are used to estimate the following: (1) U0 , the evaluation score of the product with the lowest level of each attribute. (2) For each attribute, the value of its part-worth function at each level used in the analysis. Using these information, the evaluation score of any new product concept can be estimated by interpolation. In Section 10, I show how a conjoint model is estimated using dummy variable regression. In the remainder of this section, I show how to use conjoint analysis after estimation is done.

5.2

Laptop Example

We now discuss the laptop example in more detail to show how conjoint analysis is used. In the present case, the objective of conjoint analysis is to estimate: (1) The evaluation score of the “basic” product, U0 (2) U1 (X1 ) at two or more levels in the range 4 − 10. The extreme levels 4 and 10 are always included. (3) U2 (X2 ) at two or more levels in the range 80 − 160. The extreme levels 80 and 160 are always included. Suppose we selected the following levels of the two attributes: 69

(1) Four levels of X1 : 4, 6, 8, and 10. (2) Four levels of X2 : 80, 100, 120, and 160. Several hypothetical laptop computers are created using different combinations of these attribute levels, and the respondent provides evaluation scores for these hypothetical products. For a given respondent, we obtain the following results: • U0 = mean evaluation score of the “basic laptop” with X1 = 4 and X2 = 80. • U1 (4) = 0 (by assumption) • U1 (6) • U1 (8) • U1 (10) • U2 (80) = 0 (by assumption) • U2 (100) • U2 (120) • U2 (160)

5.3

Evaluation of a New Product Concept using Linear Interpolation

New product concepts need not have attributes exactly at the levels used in conjoint estimation. For such intermediate levels, we use linear approximation to estimate part-worth utilities. We illustrate this with an example. Example: Suppose, for a given individual, we have the following estimates: • U0 = 60 • U1 (4) = 0 • U1 (6) = −10 • U1 (8) = −20 • U1 (10) = −40 • U2 (80) = 0 70

• U2 (100) = 20 • U2 (120) = 40 • U2 (160) = 50 We can express the part-worth functions graphically as follows: U1 (X1 ) 0

6

4

6

8

10

X1

-

−10

−20

−40

U2 (X2 ) 6 50 40

20

0

80

100

-

120

160 X2

71

Note that we know U1 (X1 ) only at X1 = 4, 6, 8, and 10, and U2 (X2 ) only at X2 = 80, 100, 120, and 160. To approximate the curves, we join successive points by straight lines. This is called linear interpolation. We now get the following graphs: U1 (X1 ) 0

6

4

6

8

10

X1

-

HH

HH

HH

HH

HH

−10

HH

HH

HH

HH

HH

−20

@

@

@

@

@ @

@

@ @

−40

U2 (X2 ) 50

@

6 »»

» »»» »»»

»»»

40

¡

»» »»

¡

¡

¡

¡

¡

¡

20

¡

¡

¡

¡

¡

¡

¡

0

¡

¡

80

100

-

120

160 X2

With these graphs, we can estimate the evaluation score of any laptop with X1 in the range 410, and X2 in the range 80-160. We should not use these results to evaluate product concepts with attribute levels outside the ranges used to estimate the model. 72

Consider two new product concepts: Concept 1: X1 = 5.5, X2 = 90 Concept 2: X1 = 9, X2 = 150 We can evaluate the two concepts as follows:

U0

U1 (X1 )

Concept 1 (X1 = 5.5, X2 = 90)

Concept 2 (X1 = 9, X2 = 150)

60

60

n 5.5 − 4 o n o n 9−8 o n o U1 (4) + + × U1 (6) − U1 (4) U1 (8) + × U1 (10) − U1 (8) 6−4 10 − 8 n 1.5 on o 1 =0+ (−10) − 0 = −7.5 = −20 + × {(−40) − (−20)} = −30 2 2

U2 (X2 )

U2 (80) n 90 − 80 o + × {U2 (100) − U2 (80)} 100 − 80 10 =0+ × (20 − 0) = 10 20

U2 (120) n 150 − 120 o + × {U2 (160) − U2 (120)} 160 − 120 30 = 40 + × (50 − 40) = 47.5 40

U (X1 , X2 )

60 − 7.5 + 10 = 62.5

60 − 30 + 47.5 = 77.5

Therefore, this person will prefer concept 2 to concept 1. Note: You can compute the utilities graphically also: • X1 = 5.5 is 1.5 above 4, that is, it is 3/4 th of the way from 4 to 6. Thus, U1 (5.5) is three-fourths of the way from 0 to −10, that is, −7.5. • X1 = 9 is half-way between 8 and 10. Thus, U1 (9) is half-way between −20 and −40, that is, −30. • X2 = 90 is half-way between 80 and 100. Thus, U2 (90) is half-way between 0 and 20, that is, 10. • X2 = 150 is three-fourths of the way from 120 to 160. Therefore, U2 (150) is three-fourths of the way from 40 to 50, that is, 47.5. When you have several new product concepts, you can proceed this way to identify the concept this person would prefer. 73

5.4

Prediction of Market-Share

In practice, we have parameter estimates for a representative sample of respondents. For each member of the sample, we can predict if a given new product concept will be preferred to existing products in the market. By aggregating over the sample, we can predict the market share of the new product concept.

5.5

Example: Credit Card

In Fall 2001, I conducted conjoint analysis for an MBA Marketing Management class (MAR 600) at Syracuse University. The conjoint survey, provided in the Appendix to this section, included two conjoint questionnaires, one on credit cards, and the other on personal computers. While the attribute levels for computers are out of date, those for credit cards are not. I discuss the credit card survey and results here, and the personal computer survey and results in the next section. 5.5.1 Model and analysis: Hypothetical credit cards were designed using 4 levels each of three attributes: (1) X1 (interest rate): 6, 9, 12, or 18 percent APR. (2) X2 : (credit limit): 5, 10, 25, or 50 (unit=$1000). (3) X3 (annual fee): 0, 10, 20, or 50. We assumed that the utility for a product (X1 , X2 , X3 ) can be expressed as: U (X1 , X2 , X3 ) = U0 + U1 (X1 ) + U2 (X2 ) + U3 (X3 ), where: • U0 = rating of credit card with X1 = 6, X2 = 5, and X3 = 0 • U1 (6) = 0 • U2 (5) = 0 • U3 (0) = 0 U1 (X1 ), U2 (X2 ) and U3 (X3 ) are marginal contributions of the three attributes to utility as their levels increased from the lowest points. The respondents ranked 18 hypothetical credit cards, given in the table below, from best (1) to worst (18). 74

Card # Interest Rate Credit Limit Annual Fee Rank 1 6% $5000 $0 2 12% $25,000 $10 3 6% $50,000 $0 4 12% $5000 $50 5 18% $50,000 $0 6 6% $25,000 $20 7 18% $5000 $10 8 12% $50,000 $20 9 18% $5000 $50 10 9% $10,000 $50 11 18% $10,000 $20 12 9% $50,000 $10 13 12% $10,000 $0 14 18% $25,000 $50 15 6% $10,000 $10 16 9% $5000 $20 17 6% $50,000 $50 18 9% $25,000 $0 Credit cards 3 (best on all attributes) and 9 (worst on all attributes) served as anchors. Every respondent should have given card 3 a rank of 1, and card 9 a rank of 18. 5.5.2 Data and Estimation: Data from 66 respondents (MBA students) were analyzed. For each respondent, data for cards 3 and 9, inserted as anchors, were dropped. The other 16 cards were re-ranked from 1 (best) to 16 (worst). Then, this rank was subtracted from 17 to get a score on a 1-16 scale (1: lowest, 16: highest). Using these scores, we obtained the following from each respondent: • U0 = the rating of the basic credit card with X1 = 6, X2 = 5, and X3 = 0 on a 1-16 scale. (Denoted by U 0) • U1 (6) = 0 (Denoted by U 11) • U1 (9): The marginal change in utility if X1 is 9 instead of 6. (Denoted by U 12) • U1 (12): The marginal change in utility if X1 is 12 instead of 6. (Denoted by U 13) • U1 (18): The marginal change in utility if X1 is 18 instead of 6. (Denoted by U 14) • U2 (5) = 0 (Denoted by U 21) • U2 (10): The marginal change in utility if X2 is 10 instead of 5. (Denoted by U 22) • U2 (25): The marginal change in utility if X2 is 25 instead of 5. (Denoted by U 23) • U2 (50): The marginal change in utility if X2 is 50 instead of 5. (Denoted by U 24) 75

• U3 (0) = 0 (Denoted by U 31) • U3 (10): The marginal change in utility if X3 is 10 instead of 0. (Denoted by U 32) • U3 (20): The marginal change in utility if X3 is 20 instead of 0. (Denoted by U 33) • U3 (50): The marginal change in utility if X3 is 50 instead of 0. (Denoted by U 34) 5.5.3 Evaluating a new product idea: For any given respondent, the utility of a product concept with attribute values within the ranges used in estimation can be estimated as follows: First get the components: (1) U0 (2) U1 (X1 ): Compute it as follows: X1 − 6 6 ≤ X1 < 9: U1 (X1 ) = U 12 ∗ ( ). 3

X1 − 9 ). 3 X1 − 12 12 ≤ X1 ≤ 18: U1 (X1 ) = U 13 + (U 14 − U 13) ∗ ( ). 6 (3) U2 (X2 ): Express X2 in units of $1000, e.g., use 15 if credit limit is $15,000. Compute U2 (X2 ) as follows: X2 − 5 5 ≤ X2 < 10: U2 (X2 ) = U 22 ∗ ( ). 5 X2 − 10 ). 10 ≤ X2 < 25: U2 (X2 ) = U 22 + (U 23 − U 22) ∗ ( 15 X2 − 25 25 ≤ X2 ≤ 50: U2 (X2 ) = U 23 + (U 24 − U 23) ∗ ( ). 25 (4) U3 (X3 ): Compute this as follows: X3 0 ≤ X3 < 10: U3 (X3 ) = U 32 ∗ ( ). 10 X3 − 10 10 ≤ X3 < 20: U3 (X3 ) = U 32 + (U 33 − U 32) ∗ ( ). 10 X3 − 20 ). 20 ≤ X3 ≤ 50: U3 (X3 ) = U 33 + (U 34 − U 33) ∗ ( 30 Then add: U (X1 , X2 , X3 ) = U0 + U1 (X1 ) + U2 (X2 ) + U3 (X3 ). 9 ≤ X1 < 12: U1 (X1 ) = U 12 + (U 13 − U 12) ∗ (

When you compare multiple new product concepts, a given person will select the option that yields the highest utility. 5.5.4 Benefit Segments using Cluster Analysis: I performed cluster analysis using SPSS. Each of the 66 cases in the data set is expressed in terms of 16 coordinates: the scores of the 16 profiles used (that is, 17−rank), denoted by S1 , S2 , . . ., S16 . The program computes the distances between pairs of cases in this sixteen-dimensional space and tries to create clusters based on proximity. The user specifies the number of clusters to be used. I tried both a twocluster solution and a three-cluster solution. The three-cluster solution was almost the same as the two-cluster solution with only 4 cases in the third cluster. Therefore, I only provide 76

the two-cluster results. For each cluster, the average values of U0 and the part-worth scores are given below. Cluster 1: U0 = 15.3429 U1 (6) = 0 U2 (5) = 0 U3 (0) = 0

U1 (9) = −3.0071 U2 (10) = 0 U3 (10) = −1.1857

U1 (12) = −6.1857 U2 (25) = −0.6143 U3 (20) = −1.8071

U1 (18) = −10.2643 U2 (50) = −1.5571 U3 (50) = −2.7500

U1 (9) = −1.1048 U2 (10) = 1.4032 U3 (10) = −2.8629

U1 (12) = −2.0645 U2 (25) = 2.5323 U3 (20) = −5.5887

U1 (18) = −3.3145 U2 (50) = 2.9032 U3 (50) = −9.0323

Cluster 2: U0 = 12.7823 U1 (6) = 0 U2 (5) = 0 U3 (0) = 0

Thus, it appears that Cluster 1 is largely concerned with interest rate, and Cluster 2 is largely concerned with annual fee. I don’t have a clear explanation for the negative part-worths of credit limit for Cluster 1. It may be simply random error, given that credit limit is not important anyway. It may also be worry about identity theft or misuse of a high credit limit. 5.5.5 Output: If you are interested, you may work with the Excel output file creditf01.xls posted as a Blackboard course document. For each of the 66 observations, you have the following data: Case number (1 through 66) Gender: 1 if male, 0 if female Age: 1-5 scale NCARD: Number of credit cards owned CLUS: Cluster membership, 1 if cluster 1 and 2 if cluster 2. S1 − S16 : The scores of the sixteen credit cards used in the analysis. Conjoint Estimates: U 0, U 11, U 12, U 13, U 14, U 21, U 22, U 23, U 24, U 31, U 32, U 33, U 34.

5.6

Example: Personal Computer

We used the data from the MBA class in Fall 2001 to do a second conjoint analysis on personal computers. The survey is included in the Appendix to this section. 5.6.1 Model and analysis: The hypothetical products were designed using 4 levels each of three attributes: (1) X1 (warranty length): 1, 2, 3, or 4 years. (2) X2 : (RAM): 1, 2, 3, or 4 (unit= 128 MB). (3) X3 (hard drive size): 1, 2, 3, or 5 (unit = 20 GB). 77

We assumed that the utility for a product (X1 , X2 , X3 ) can be expressed as: U (X1 , X2 , X3 ) = U0 + U1 (X1 ) + U2 (X2 ) + U3 (X3 ), where: • U0 = utility for personal computer with X1 = 1, X2 = 1, and X3 = 1 • U1 (1) = 0 • U2 (1) = 0 • U3 (1) = 0 U1 (X1 ), U2 (X2 ) and U3 (X3 ) are marginal contributions of the three attributes to utility as their levels increased from the lowest. Each respondent rated 18 hypothetical personal computers listed in the table below from 1 (best) to 18 (worst). Data from computers 1 and 6, inserted as anchors, were dropped. The remaining 16 products were re-ranked from 1 to 16. Each PC was then given a score of (17−rank), that is, the PC ranked 1 got a score of 16, the PC ranked 2 got a score of 15, so on. Personal Computer # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Warranty

4 1 2 3 4 1 1 2 3 4 1 2 3 4 1 2 3 4

yr yr yr yr yr yr yr yr yr yr yr yr yr yr yr yr yr yr

Random Access Memory (RAM) (in MB) 512 128 256 384 512 128 256 384 512 128 384 512 128 256 512 128 256 384

Hard Drive Size (in GB) 100 60 20 60 20 20 100 40 100 40 20 60 20 60 40 100 40 100

Rank

5.6.2 Data and Estimation: Data from 68 respondents (MBA students) were analyzed. For each respondent, data for computers 1 and 6, inserted as anchors, were dropped. The other 16 78

computers were re-ranked from 1 (best) to 16 (worst). Then, this rank was subtracted from 17 to get a score on a 1-16 scale (1: lowest, 16: highest). Using these scores, we obtained the following from each respondent: • U0 = the rating of the basic PC with X1 = 1, X2 = 1, and X3 = 1 on a 1-16 scale. (Denoted by U 0) • U1 (1) = 0 (Denoted by U 11) • U1 (2): The marginal change in utility if X1 is 2 instead of 1. (Denoted by U 12) • U1 (3): The marginal change in utility if X1 is 3 instead of 1. (Denoted by U 13) • U1 (4): The marginal change in utility if X1 is 4 instead of 1. (Denoted by U 14) • U2 (1) = 0 (Denoted by U 21) • U2 (2): The marginal change in utility if X2 is 2 instead of 1. (Denoted by U 22) • U2 (3): The marginal change in utility if X2 is 3 instead of 1. (Denoted by U 23) • U2 (4): The marginal change in utility if X2 is 4 instead of 1. (Denoted by U 24) • U3 (1) = 0 (Denoted by U 31) • U3 (2): The marginal change in utility if X3 is 2 instead of 1. (Denoted by U 32) • U3 (3): The marginal change in utility if X3 is 3 instead of 1. (Denoted by U 33) • U3 (5): The marginal change in utility if X3 is 5 instead of 1. (Denoted by U 34) 5.6.3 Evaluating a new product idea: For a given respondent, the utility of a product concept with attribute values within the ranges used in estimation can be estimated as follows: First obtain the components: (1) U0 (2) U1 (X1 ): Express X1 in years. Compute U1 (X1 ) as follows: 1 ≤ X1 < 2: U1 (X1 ) = U 12 ∗ (X1 − 1). 2 ≤ X1 < 3: U1 (X1 ) = U 12 + (U 13 − U 12) ∗ (X1 − 2). 3 ≤ X1 ≤ 4: U1 (X1 ) = U 13 + (U 14 − U 13) ∗ (X1 − 3). (3) U2 (X2 ): Express X2 in units of 128 MB, e.g., use 1.5 if RAM is 192 MB. Compute U2 (X2 ) as follows: 1 ≤ X2 < 2: U2 (X2 ) = U 22 ∗ (X2 − 1). 2 ≤ X2 < 3: U2 (X2 ) = U 22 + (U 23 − U 22) ∗ (X2 − 2). 3 ≤ X2 ≤ 4: U2 (X2 ) = U 23 + (U 24 − U 23) ∗ (X2 − 3). 79

(4) U3 (X3 ): Express X3 in units of 20 GB. For example, use 2.5 if hard drive size is 50 GB. Compute U3 (X3 ) as follows: 1 ≤ X3 < 2: U3 (X3 ) = U 32 ∗ (X3 − 1). 2 ≤ X3 < 3: U3 (X3 ) = U 32 + (U 33 − U 32) ∗ (X3 − 2). X3 − 3 3 ≤ X3 ≤ 5: U3 (X3 ) = U 33 + (U 34 − U 33) ∗ ( ). 2 Then add: U (X1 , X2 , X3 ) = U0 + U1 (X1 ) + U2 (X2 ) + U3 (X3 ). When you compare multiple new product concepts, a given person will select the option that yields the highest utility. 5.6.4 Benefit Segments using Cluster Analysis: I performed cluster analysis using SPSS. Each of the 68 cases in the data set is expressed in terms of 16 coordinates: the scores of the 16 profiles used (computed as 17−rank) for the case, denoted by S1 , S2 , . . ., S16 . The program computes the distances between pairs of cases in this sixteen-dimensional space and tries to create clusters based on proximity. The user specifies the number of clusters to be used. In this case (personal computers), three distinct clusters emerged. For each cluster, the averages of U0 , U1 (1), . . . , U3 (5) are given below. Cluster 1 U0 = −0.0875 U1 (1) = 0 U1 (2) = 4.10 U2 (1) = 0 U2 (2) = 1.4125 U3 (1) = 0 U3 (2) = 0.6375

U1 (3) = 8.2125 U2 (3) = 2.8625 U3 (3) = 1.5500

U1 (4) = 10.5375 U2 (4) = 3.3250 U3 (5) = 1.7125

Cluster 2 U0 = −0.09167 U1 (1) = 0 U2 (1) = 0 U3 (1) = 0

U1 (2) = 0.5500 U2 (2) = 2.5250 U3 (2) = 3.4250

U1 (3) = 1.4417 U1 (4) = 1.5417 U2 (3) = 3.4833 U2 (4) = 4.0917 U3 (3) = 7.1417 U3 (5) = 10.1667

Cluster 3 U0 = 0.4722 U1 (1) = 0 U2 (1) = 0 U3 (1) = 0

U1 (2) = 0.6250 U2 (2) = 4.2222 U3 (2) = 1.1250

U1 (3) = 1.4722 U2 (3) = 7.7917 U3 (3) = 1.5972

U1 (4) = 1.3472 U2 (4) = 11.0417 U3 (4) = 2.8889

5.6.5 Output: The results are presented as the Excel file: pcf01.xls. For each of the 68 observations, you have the following data: Case number (1 through 68) Gender: 1 if male, 0 if female Age: 1-5 scale NCARD: Number of credit cards owned KNOW: self assessment of knowledge of PC (1-7 scale) 80

PPC: 1 if purchased PC in the last two years, 0 if not OWNPC: 1 if owns PC, 0 if not EMAIL, WORD, SURF, SPREAD, TAX, STORDAT, STAT, PROG, VIDEO, OTHER: each 1 if checked, 0 if not CLUS: Cluster membership using three-cluster solution; 1 if cluster 1, 2 if cluster 2, 3 if cluster 3. S1 − S16 : The scores of the sixteen personal computers used in the analysis. Estimates of: U 0, U 11, U 12, U 13, U 14, U 21, U 22, U 23, U 24, U 31, U 32, U 33, and U 34.

5.7

Optional Topics

5.7.1 Qualitative Attributes: To simplify the discussions, I only considered quantitative attributes such as interest rate and credit limit. However, everything we discussed applies equally well to qualitative attributes. For example, we could have used four levels of the color of an iPOD: red, blue, silver, and black. Qualitative attributes are used extensively in marketing applications. For example, researchers have used brand name and country of origin to find how these attributes affect customer evaluation. 5.7.2 Estimating Price Sensitivity: In conjoint analysis, price may be used as one of the attributes. Using the results, the researchers can estimate how customers make trade offs between an attribute level and its price. Also, for a given new product concept, it is possible to estimate its demand at a given price level. 5.7.3 Ideal Point Model: In our discussions, we did not make any assumption about what the part-worth functions U1 (X1 ), etc., look like. Also, when we estimate these functions using dummy variable regression as discussed in Section 10, we do not make assumptions about functional form. We now discuss one specific case of the additive part-worth model called the ideal point model. The ideal point model assumes that a given customer has an ideal point, (X1∗ , X2∗ , . . . , Xn∗ ), and the customer’s evaluation score of a product decreases with distance from the ideal point. The distance of a product (X1 , X2 , . . . , Xn ) from the ideal point is defined as follows: D2 = W1 (X1 − X1∗ )2 + W2 (X2 − X2∗ )2 + . . . + (Xn − Xn∗ )2 , where W1 , W2 , . . ., Wn are all greater than zero and reflect the relative importance of the different attributes. (In Assignment 2, we applied a special case of this distance where W1 = W2 = . . . = 1.) The evaluation score of any product (X1 , X2 , . . . , Xn ) can be expressed as: 2

U (X1 , X2 , . . . , Xn ) = K − D = K −

n X i=1

81

Wi (Xi − Xi∗ )2 ,

where K is a constant and equals the utility of the ideal product. Rewriting: h

U (X1 , X2 , . . . , Xn ) = K −

n X

Wi Xi∗ 2

i=1

i

+

n X

(2Wi Xi∗ Xi − Wi Xi2 )

i=1

If each Xi is expressed as the difference of the attribute level from the lowest level, then this becomes our usual additive part-worth utility model where: U0 = K −

Pn

i=1

Wi x∗i 2

U1 (X1 ) = (2W1 X1∗ )X1 − W1 X12 .. . Un (Xn ) = (2Wn Xn∗ ) − Wn Xn2 For any attribute Xi , the part-worth function is a parabola that reaches the highest point at Xi∗ as shown below. In Section 10.6 we discuss how to estimate this model. Ui (Xi ) 6

-

Xi∗ Xi 5.7.4 Developing new product concepts: To use conjoint analysis, it is necessary to develop hypothetical new product concepts. If you use commercial packages such as those from www.sawtooth.com, then the software will generate hypothetical products for you. Otherwise, you have to develop hypothetical products yourself. Note that in Sections 5.5 and 5.6, I used 4 levels each of three attributes. Thus, in each case, 4 × 4 = 64 hypothetical products were possible. Instead, we used 16 hypothetical products plus the two anchors. This was done using a method called Latin Square Design that can be used for three attributes, each attribute with the same number of levels. I now show how we can use the Latin Square design to get 16 hypothetical products when there are three attributes with four levels each. Step 1: First focus on attributes 1 and 2, and write the 4 × 4 = 16 combinations as a table: Attribute 1 1 2 3 4

1 (1,1) (2,1) (3,1) (4,1)

Attribute 2 2 3 (1,2) (1,3) (2,2) (2,3) (3,2) (3,3) (4,2) (4,3)

4 (1,4) (2,4) (3,4) (4,4) 82

Step 2: Start with row 1, and put the four levels of the third attribute sequentially. You can use any starting point. For example, you can put them as (1, 2, 3, 4), or (2, 3, 4, 1), or (3, 4, 1, 2), or, (4, 1, 2, 3). In row 2, shift them by one place to the left. For example, if you have (1, 2, 3, 4) in row 1, then it should be (2, 3, 4, 1) in row 2. Again, shift them by 1 in row 3, and shift them by 1 in row 4. Thus, depending on what you chose in row 1, you may have any of the following four Latin-Square designs: Design 1 Attribute 1 1 2 3 4 Design 2 Attribute 1 1 2 3 4 Design 3 Attribute 1 1 2 3 4 Design 4 Attribute 1 1 2 3 4

1 (1,1,1) (2,1,2) (3,1,3) (4,1,4)

Attribute 2 2 3 (1,2,2) (1,3,3) (2,2,3) (2,3,4) (3,2,4) (3,3,1) (4,2,1) (4,3,2)

4 (1,4,4) (2,4,1) (3,4,2) (4,4,3)

1 (1,1,2) (2,1,3) (3,1,4) (4,1,1)

Attribute 2 2 3 (1,2,3) (1,3,4) (2,2,4) (2,3,1) (3,2,1) (3,3,2) (4,2,2) (4,3,3)

4 (1,4,1) (2,4,2) (3,4,3) (4,4,4)

1 (1,1,3) (2,1,4) (3,1,1) (4,1,2)

Attribute 2 2 3 (1,2,4) (1,3,1) (2,2,1) (2,3,2) (3,2,2) (3,3,3) (4,2,3) (4,3,4)

4 (1,4,2) (2,4,3) (3,4,4) (4,4,1)

1 (1,1,4) (2,1,1) (3,1,2) (4,1,3)

Attribute 2 2 3 (1,2,1) (1,3,2) (2,2,2) (2,3,3) (3,2,3) (3,3,4) (4,2,4) (4,3,1)

4 (1,4,3) (2,4,4) (3,4,1) (4,4,2)

For example, in Design 4, the first product in Row 1 has level 1 of attribute 1, level 1 of attribute 2, and level 4 of attribute 4. All these designs have an important property: For each level of a given attribute is combined with each level of another attribute exactly once. Technically, the attributes have zero correlations in the design.

83

5.8

Additional Reading

1. Paul E. Green, and V. Srinivasan (1978), “Conjoint Analysis in Consumer Research: issues and outlook,” Journal of Consumer Research, volume 5, pages 103-23. 2. and (1990), “Conjoint Analysis in Marketing: new developments with implications for research and practice,” Journal of Marketing, volume 54, pages 3-19. These are excellent review papers written by two of the leading authorities on the subject. You may also visit www.sawtoothsoftware.com

5.9

Appendix: Survey for Credit Card and PC

Dear Participant: We are conducting this survey to determine what is important to you when you select credit cards and personal computers. Your response is very important to us. Any information you provide will be held strictly confidential and will not be used for any commercial purpose. Thanks for your help. Students of MBC 600, School of Management

84

Part 1: Credit Card 18 hypothetical credit cards are described on page 3 of this questionnaire (Table 1). All these credit cards are Visa cards issued by a major Midwestern bank and they all have the following standard features: • Collision and damage insurance on rental cars. • Frequent flier mileage at one mile for one dollar of purchase using the card on an airline of your choice. • 30 day grace period on payments. These cards are identical in every way except for the following: 1. Interest rate: The interest rate can be 6%, 9%, 12%, or 18% APR. 2. Credit limit: The credit limit can be $5000, $10,000, $25,000, or $50,000. 3. Annual fee: The annual fee can be $0, $10, $20, or $50. Once again, the credit cards listed on the next page are identical in every way except for the three features listed above. Assume that you are interested in getting a new credit card, and your choice is restricted to the 18 credit cards described in Table 1. Identify the one credit card in the list that best suits your personal tastes and preferences. Assign this card a rank of 1. Then, identify the card you like second best, and assign it a rank of 2. Please continue this way and rank all the 18 credit cards. Please answer carefully; your last few rankings are as important as your first few. Enter the ranks in the rank column.

85

Table 1 Please rank the following credit cards in order of preference (from 1: best to 18: worst). All these credit cards are identical in every way except for interest rate, credit limit, and annual fee. Card # Interest Rate Credit Limit Annual Fee Rank 1 6% $5000 $0 2 12% $25,000 $10 3 6% $50,000 $0 4 12% $5000 $50 5 18% $50,000 $0 6 6% $25,000 $20 7 18% $5000 $10 8 12% $50,000 $20 9 18% $5000 $50 10 9% $10,000 $50 11 18% $10,000 $20 12 9% $50,000 $10 13 12% $10,000 $0 14 18% $25,000 $50 15 6% $10,000 $10 16 9% $5000 $20 17 6% $50,000 $50 18 9% $25,000 $0

86

Part 2: Personal Computer 18 hypothetical personal computers are described in Table 2 of this questionnaire. All these personal computers are offered by companies with similar reputation at the same price, $1500, and they all have the following standard features: • 17” Sony Trinitron monitor • 1.8 GHz Intel Pentium 4 Processor • Ultra ATA hard drive, 7200 RPM • 16x max DVD Rom • 12x max CD-RW drive • 3.5” floppy drive and 250 MB Zip drive • Microsoft Windows XP Home edition with Microsoft Works Suite 2001 • 64 MB NVidia GeForce 2 MX graphics card • Harmon Kardon HK695 surround sound speakers with subwoofer • Sounblaster Pro 16 soundcard • 56 K modem and 10/100 network card • i.Link I-EEE 1394 interface • HP 845C color printer • free 12 month internet access by MSN However, these personal computers differ on the following three features: hard drive size, random access memory (RAM), and term of warranty, as described on the next page.

87

The personal computers can differ on the following three features: (1) Hard drive size: The hard drive size can be 20 GB, 40 GB, 60 GB, or 100 GB. (2) Random Access Memory (RAM): (133 MHz RDRAM) The random access memory can be 128 MB, 256 MB, 384 MB, or 512 MB. (3) Warranty: The warranty can be: 1 yr: One year limited warranty with next business day on-site service; one year online help. 2 yr: Two year limited warranty with next business day on-site service; two year online help. 3 yr: Three year limited warranty with next business day on-site service; lifetime online help. 4 yr: Four year limited warranty with next business day on-site service; lifetime online help. Assume that you are interested in buying a personal computer for personal use, and your choice is restricted to the 18 personal computers described in Table 2. Please rank these 18 personal computers in order of preference. Assign a rank of 1 to the personal computer that best suits your personal tastes and preferences. Then find the next best personal computer for yourself, and assign it a rank of 2. Proceed this way until you have ranked all 18 personal computers. Please answer carefully; your last few rankings are as important as your first few.

88

Table 2 Please rank the following personal computers in order of preference (from 1:best to 18: worst). All personal computers have the same price ($1500), and are identical in every way except for warranty, random access memory, and hard-drive size. Personal Computer # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Warranty

4 1 2 3 4 1 1 2 3 4 1 2 3 4 1 2 3 4

yr yr yr yr yr yr yr yr yr yr yr yr yr yr yr yr yr yr

Random Access Memory (RAM) (in MB) 512 128 256 384 512 128 256 384 512 128 384 512 128 256 512 128 256 384

89

Hard Drive Size (in GB) 100 60 20 60 20 20 100 40 100 40 20 60 20 60 40 100 40 100

Rank

Part 3: background Information 1. Gender:

Male

Female

2. Age: 20 or below over 60

21-25

26-40

41-60

3. How many credit cards do you have at present? credit cards 4. Compared to an average person, how knowledgeable are you about personal computers? (Please circle) Much less knowledgeable 1 2 3

4

Much more knowledgeable 5 6 7

5. Did you purchase a personal computer within the last two years? Yes No 6. Do you currently own a personal computer? Yes No 7. If you do currently own a personal computer, what have you used it for during the last three months? (Check all that apply. If you do not currently own a personal computer, please skip the question.) (a)

send/receive email

(b)

word processing

(c)

(d)

prepare spreadsheets

(e)

prepare tax forms (f)

store data

(g)

run statistical packages (h)

do programming

play video games

(j)

other

(i)

**Thank You for Your Participation**

90

surf the internet

6 6.1

Selected Topics on Product Strategy Introduction

A “product” is everything, favorable and unfavorable, that a person receives in an exchange. A product is a very general term. It may be a pure good (e.g., a pair of blue jeans), a pure service (e.g., an airplane ride to New Delhi), or a combination of both (e.g., a dinner at an expensive restaurant, or a personal computer with on site service for one year). You may often have a situation when the same need may be satisfied by a good and a service. For example, your entertainment need may be satisfied by going to a movie theater (a service) or buying a video tape (a good) and watching it at home. Similarly, one may decide to use disposable diapers (a good) or a cloth diaper service. It is useful to think of a product at three levels (the Kotler/Keller text adds two more levels). Core product: Basic benefit or service provided. For example, it may be the chemical acetaminophen. Actual Product: Core product plus brand name, packaging, features, style. With these, acetaminophen becomes Tylenol, Panadol, or a store brand. Augmented Product: Actual product + support (e.g., warranty, service, financing, delivery, etc.) + price. Example of augmented products: Singer, a manufacturer of sewing machines, provides sewing lessons, the car manufacturer Acura provides 24 hour roadside service to its US customers. Note: Even when the core product is the same, a marketer can differentiate the augmented product through pricing, distribution, customer service, and even the effective use of promotion to create the right image for the product. For example, Duracell has created an image of a long lasting tough product. Marlboro has an image of rugged independence.

6.2

Classification of Consumer Products

Consumer products can be divided into four main categories, (1) convenience products, (2) shopping products, (3) specialty products, and (4) unsought products. We will discuss them separately. 6.2.1 Convenience products: These can be subdivided into three categories, staple products, emergency products and impulse products. (a) Staple products: These are frequently purchased items, typically necessities of life, e.g., milk, bread, newspaper, car wash, dry cleaning etc., usually commodities with standard prices. Key Feature: These are usually commodities, i.e., the consumer finds little difference in the offerings of different manufacturers. If a particular brand is not available at a store, a consumer 91

is unlikely to go from store to store to find it. (b) Emergency products: These are products that consumers need immediately. Examples include headache medicine, snow shovels, ambulance, the hospital emergency room, taxi to get back home when the car breaks down, American Automobile Association (AAA) service to tow your car. Key feature: Once again, when the need arises, the consumer needs the product immediately and is not able to shop around. For example, if your car is about to run out of gas on a lonely highway, you would be unlikely to shop around for the best price. (c) Impulse products: These are products one does not plan to purchase and decides to buy only when she see them. Examples include magazines, candy, flowers, etc. Retailers usually place many of these items near the check-out counter to increase the chance that they will be noticed and hence purchased. The examples I have given so far are all goods, but impulse purchase may include services also. For example, you may pass by a beauty salon and decide to have your fingernails manicured, or see the phone in front of your seat during an airline flight and proceed to call home. Key feature: An impulse product is not purchased if the consumer does not notice it. Thus, impulse products are packaged and placed to attract attention. Common features of convenience products: a) Consumers generally have adequate knowledge of the product category. b) The consumer does not derive pleasure from the purchase activity itself, and does not usually shop around. This makes product availability very important. The marketer’s perspective: • The product must be widely available. The manufacturer therefore has to distribute the product through many retail outlets. • The retailer usually does not advertise the product since that advertisement will benefit all retailers and not just him. As a result, the manufacturer must bear the cost of promoting the product to the ultimate consumer. • Due to the size of the target audience, promotion is usually done through advertising and the use of coupons. Companies like P & G and Johnson & Johnson, who market branded products, do a lot of advertising about the specific brands they market. The marketers of commodities often join together and advertise their common, generic, product in order to generate primary demand; examples include the National Pork Producers sponsoring pork commercials, and Florida citrus growers airing commercials about Florida orange juice. • The only time retailers promote a convenience product is when they group them together 92

and promote them simultaneously (e.g., using coupons inserted in Sunday newspapers) to attract customers to the store. Note: Whether a product is a staple, an emergency product or an impulse product may depend on the specific buying situation encountered by the consumer. For example, a staple like gasoline may become an emergency product if you are stranded on a deserted highway. Also, these days people have less time to spare and they expect to find everything they need when they go to a supermarket. As a result, they often do not plan what they will purchase when they go to grocery stores. Recent studies show that over 50 % of all supermarket purchases are now “impulse” purchases, i.e., purchases without any prior planning. This makes availability of product as well as attractive package design and shelf location (to attract attention) very important. 6.2.2 Shopping products: Shopping products are characterized by the fact that consumers perceive considerable difference in the augmented product and a purchase represents a greater monetary or emotional investment compared to the purchase of convenience products. As a result, the consumers shop around and compare price, quality, brands, etc., and many people derive pleasure from the “shopping around” itself. Note that a customer shops around only when she sees a benefit in doing so. While this will occur for major purchases like that of a car or a house, it also occurs for purchases that do not cost much but have important consequences. For example, a person may look around to find a reliable day care provider. Also, if the product use is publicly observable (e.g., a suit, an evening dress, a hair cut), the consumer is likely to be choosy. When a consumer shops around, she may not necessarily buy a product right now. She may simply be gathering information which she may use later. Even when she is ready to make a purchase, she has usually made a decision about the product category, but not a specific brand. Categories of shopping products: Shopping products may be placed in two categories based on the nature of the augmented product: (i) Heterogeneous Shopping Products: Here, the actual product itself varies considerably, and the consumer shops around until she finds one which she really likes. For example, when you are trying to buy a jacket, you may have to look around until you find one which fits you perfectly. (ii) Homogeneous Shopping Products: Many products such as appliances (e.g., vacuum cleaner, refrigerator, sewing machine, microwave oven) are standardized products, but the price (and availability) may vary considerably from store to store. Here also the consumer is likely to shop around for the best price. Issues Regarding Shopping Products: (a) The manufacturer’s point of view: The manufacturer usually limits distribution, especially if the product has a high quality image (e.g., Sony TV or IBM personal computer). 93

This has the following advantages: 1. The manufacturer is not hurt by limited distribution since the consumer will look around. 2. The cost of distribution is lower as a smaller number of retail outlets need to served. 3. As there are fewer retailers selling the product, there is a smaller chance that they will engage in price competition. [Price competition among retailers lowers the price the consumer pays, and it eventually lowers the profit of the manufacturer as he must now sell the product at a lower price to all retailers.] 4. Limiting distribution allows the manufacturer to be selective. For example, a marketer of a high quality product like Sony TV may like to avoid retailers who will not provide adequate customer service or engage in loss leader pricing, i.e., selling the product at or below cost to attract customers to the store (this will result in intense price competition among retailers and hurt the manufacturer). For example, Compaq, a maker of high quality personal computers, at one time only distributed its products through retailers who have salespeople with special training about Compaq products. 5. If the manufacturer tries to get wider distribution, higher quality retailers may drop the product. (b) The importance of the retailer: When a shopping product is involved, a consumer has usually made a decision about the product category but not the specific product. She is therefore likely to choose retail outlets to visit where she thinks she will find the product she wants. The manufacturer must therefore place its product at the right kind of outlet, and store image becomes very important. (c) The sharing of promotional costs: Due to limited distribution, the retailer directly benefits from promotion. Thus, promotional costs are often shared between the manufacturer and the retailer. This mechanism has a name: co-operative advertising (also called co-op advertising). (d) Formation of clusters of retailers: The retailer of shopping products (in particular heterogeneous products) recognizes that it must provide the consumer enough choice. Sometimes small retailers accomplish that by forming a cluster where the consumer will find enough variety. You observe this phenomenon in shopping malls where retailers of apparel are often located next to one another. Here, the retailers forming the cluster are jointly offering a prospective customer a reason to visit that location. 6.2.3 Specialty products: Examples include collectors’ items, products with unique features etc. 94

Key characteristic: The consumer thinks that the product is unique and therefore does not consider a substitute acceptable. As a result, price and availability are not very important and consumers will wait for a specific product. For example, a consumer who has made a decision to attend a performance by the New York Metropolitan Opera is unlikely to go to concert by Britney Spears instead. Note: A specialty product does not always have to be costly. For example, a person may drive a long distance to buy muffins from a bakery which she really likes. 6.2.4 Unsought products: These fall in two categories: (a) New products that the consumer is not yet aware of. Most new products start out as unsought products because the potential customers do not know the benefits of using the product. For example, customers were slow to accept products like instant coffee and main frame computers. In this case, the marketers must educate the potential customers about potential benefits. IBM had to perform extensive personal selling to show businesses how a main frame computer can improve business productivity. (b) Products that consumers do not want to think about. Examples include funeral plots and life insurance. As the selling task is difficult, businesses often have to work hard to motivate their salespeople. For example, insurance companies sometimes pay up to a 60% commission rate for the sales of whole life insurance policies.

6.3

Branding

6.3.1 Introduction: A brand is a name, term, symbol, design, or a combination that identifies a seller’s products and differentiates them from competitors’ products. The following terms are useful. Brand-name: This is the part of a brand that can be spoken, including letters, words, and numbers. It identifies a product or a group of products and sometimes applies to a product line, e.g., Honda Civic, Accord, Prelude. Objective of Branding: Differentiate your product. Make it more than a commodity (e.g., Gold Medal flour, Morton salt, Weaver chicken). Importance of branding: 1. Source of information for consumers: reduces information search by acting as surrogate for detailed attribute information. 2. Makes it easy for consumers to identify goods or services. 3. Helps differentiate product from competitors’ products. This may reduce price sensitivity of the consumer. 4. Branding can be helpful when introducing a new product as it reduces market development 95

costs by reducing the cost of promotion and distribution. For instance, that is extremely useful for short lived products like designer clothes and toys. Note: To use branding effectively, the marketer should be able to promote the brand and also maintain a consistent quality of output. When these are not possible, branding may not be used. 6.3.2 Family versus Individual Branding A) Family Branding: Grouping many products under one brand name, e.g., Singer sewing machines, Frigidaire appliances, Revlon beauty products, or Fisher Price toys. This is very useful when the product line is constantly changing and any given product has a short life span (e.g. designer clothes). Family branding may come at three levels: (1) Same family name for all products offered by the company. Examples include Fisher Price toys, Liz Claybourne fashion clothes, and Revlon cosmetics. What do these products have in common? (2) Family name plus an individual name, such as Toyota Camry and Honda Civic. The family name conveys the shared properties of the brands (such as reliability) while the individual name signifies specific aspects of a given brand (such as size). This strategy is used extensively by car manufacturers. (3) Groups of products, each with a different family name. For example, the US retail giant SEARS uses the Kenmore name for its “white” appliances (refrigerator, washing machine, etc.), and the Kraftsman name for its tools. Possible benefits of family branding: 1. Advertising and promotion may help all products. 2. New product introduction is easier to accomplish. Possible problems of family branding: 1. Any negative publicity (such as product failure) is also shared by all products with the family name. 2. Need to make sure the family name has a consistent meaning. If you give the same family name to two totally different products with nothing in common (such as: perfume and chewing tobacco, or health food and cigarettes), the family name will not convey any useful information to the customer. 3. If you market competing products, the customer may think they are close substitutes. This may result in cannibalization, that is, your products eating up each other’s market share. 96

Usually, a new product is included in the “family” if any of the following holds: 1) The new product is a small variation of an old product, or, even if that is not true, it is closely related to the family (e.g., GM auto parts are related to the GM cars). 2) The introduction of the new product may enhance the image of the entire family. For example, the introduction of the Athlon chip which performed better than earlier versions of Intel Pentium chip has established AMD as a strong contender in the chip market. In addition, the new product is typically given the family name when there is little risk of damaging the family name from doing that (e.g., when a new medicine results in deaths). B) Individual Branding: Using different brand names for different products. This is clearly more costly than family branding as each product needs to be promoted individually. When is that justified? Some reasons to use it: 1. Avoid cannibalization: Less chance that consumers will think that the new brand is a ‘new improved’ version of an old brand. Better chance that users of competitors’ brands will switch to the new product. This is particularly relevant when your company offers several products to the same market segment, and that may be the reason why P & G uses individual branding exclusively. 2. To make sure the failure of one product will not harm other products. 3. The products are very different from one another or are aimed at very different customer groups. Thus, no consistent image will be presented by using the same brand name (e.g., if Skoal, the makers of chewing tobacco, decides to make expensive perfumes). Parker, the makers of writing equipment, had this problem as it marketed pens ranging from $2/unit “jotters” to fountain pens with price of over $500. Thus, just the name Parker had little meaning to a consumer. Levi Strauss, the makers of blue jeans, also markets a line of expensive formal clothing called “Oxford.” It never promotes the fact that these products are marketed by the same entity. 6.3.3 Brand Loyalty: You come across this term frequently when you read the marketing literature. Brand loyalty is defined as a consistent preference for one brand over others. Basically, it is the tendency of a customer to intentionally purchase a product repeatedly. (Similar concepts: Store loyalty, product category loyalty). 97

Some reasons why brand loyalty may exist: 1. Trying a new brand may involve risk. For example, people often do not like to switch medicines they take for headaches or cold. Tylenol uses this fear as the basis of its ad theme: “Recommended by physicians.” 2. Sunk investment plus alternatives are perceived to be similar. In this case, consumers see little reason to make the effort needed to switch. For example, people rarely switch banks. As a result, banks promote heavily to new arrivals in town (e.g., newly recruited professors at SU) who have not yet attached themselves to particular banks. 3. Strong emotional involvement e.g., Coke, political party, school. There is no universally accepted method to quantify brand loyalty. Some of the alternatives marketers use to quantify brand loyalty are as follows: 1. Analyze pattern of purchases relative to competition. The objective here is to determine the probability that a person will purchase a brand in the future. The probability is the measure of brand loyalty. For example, if it equals 1, we know that the person will never switch. A crude measure of this probability is the percentage of times the brand is purchased, which can be computed easily from purchase data. 2. Ask a customer how much more money he would pay for a given brand compared an alternative, or, equivalently, how much price cut by the competition would make him switch (this is called the “dollar-metric”). This is particularly useful to a marketer as it tells him how much of a price cut will be needed to induce a product switch. Comparison: Method 1 uses data on past purchases alone and cannot measure underlying likes/dislikes of a consumer. In contrast, method 2 directly questions respondents, and can suffer from biased answers. Note: Brand loyalty can be a basis for consumer segmentation as marketers often use sales, coupons etc. to attract the switching segment.

6.4

Product Life Cycle

Terminology Primary demand: demand for product category. Selective demand: demand for a specific brand within the category. The four stages of the life cycle listed below hold for a product class. We now consider the transition of a new product through the four stages of its life cycle. 1. Introduction/Market Development Characteristic: Most new products start out as unsought products as the consumers have not yet grasped the nature of the benefit it will provide. 98

−→ Consumer awareness is low. Competitors are waiting. Pioneer’s objectives: Get consumers to try product and stores to carry it −→ Need to generate primary demand for the product. Need to promote the product to both consumers and channels of distribution. Need to develop proper marketing mix with strategic goals. Length of Stage: Depends on (i) complexity, (ii) perceived risk due to newness, (iii) ability to satisfy consumer needs, and (iv) presence of substitutes and the level of consumers’ satisfaction with the substitutes. Some courses of action: • A small number of innovators will try it first. It is crucial that they like it −→ The marketer must make sure that the product is used properly. For example, if it is an appliance, technical support should be provided. Any problems should be corrected quickly. Strict quality control is needed, specially if the product is heavily promoted (a big reason for the failure of Edsel was due to the fact that it was so heavily promoted that its quality problems also received a lot of publicity). • At this stage, heavy promotion is usually needed directed at both resellers and the ultimate customers of the product. Sometimes it makes sense to cooperate with other organizations to develop the market (e.g., RCA shared its technology with other manufacturers to help develop the market for color television sets). • Need to provide incentives to dealers to carry the product and promote it if necessary. This strategy of providing incentives to marketing intermediaries (e.g., retailers) is called a push strategy. • The specific nature of promotion that should be adopted in the introductory stage depends on the nature of the product. For products with a high unit price, personal selling with a knowledgeable salesforce is appropriate. IBM used this method to educate potential customers about the benefits of using a main frame computer. For medium priced appliances, dealers are often offered incentives to promote the product to the consumers. For low unit price consumer products, coupons and mass advertising are effective mechanisms to induce trial. Sales: During the introductory stage, sales are slow. If the product is successful, there is a gradual rise of sales. Profits: Usually negative due to heavy costs of research & development and promotion to both consumers and channels of distribution. Lack of experience and inability to have economy of scale drives up cost of production. 2. Growth 99

Basic Characteristic: At this stage, many consumers are aware of the product, and as they want the product, retailers carry it in self interest. In marketing terminology, we say that demand now pulls the product. The product is now widely available. At this stage, competitors are assured of the demand for the product category and enter the market. As a result, the focus of promotion shifts to generating selective demand, i.e., the demand for your specific product. Pioneer’s problem: Get consumers to prefer product compared to what the competition offers in order to reduce competitive pressure. Usual Strategy of pioneer: Differentiate product. This reduces the need for price competition. Note: While it is usually possible to avoid head on competition by differentiating product, there are cases when you cannot avoid it even during growth. That happened in the market for VCR’s where the format (VHS) with the larger market share became the industry standard. Sales: Accelerates, i.e., the rate of increase itself increases. The experience curve phenomenon and economy of scale (see Sections 6.6 and 6.8 of the Reader) jointly reduce the cost of production and distribution. Profit: Increases at first, then tends to be flat as prices come down to reflect the increase in competition. 3. Maturity Basic characteristics: High consumer awareness. Intense competition, including generic products. As a result, the margin of profit is low and marginal competitors drop out (the so called “shake out” phenomenon). Strategy : Product differentiation and heavy promotion to support it. Price competition can no longer be avoided, and we observe a high degree of tactical price competition in the form of temporary price cuts, coupons, etc. Sales: Demand for product category is now at its highest point. Costs: The marginal cost of production is at the lowest point. Profit: Due to competition, prices drop close to the marginal cost of producing and selling the product. Due to the lowered profit margins, profits gradually decline. 4. Decline Basic Characteristics: Demand declines, usually due to the introduction of a superior substitute. Intense price competition. More competitors drop out. Some Strategies for the product line: • Retrenchment: This is similar to concentrated marketing. The marketer drops some products and focuses on specific markets. 100

• Harvesting: The marketer does not reinvest in the product. The revenue generated is invested in other businesses. This is a phased withdrawal strategy. • Consolidate: This is adopted by the stronger competitors. The idea is to capture the small market as competitors drop out. Enlightened Strategy (during maturity or decline): Extend Product Life Cycle • Find new users for product. • Find new uses for product. • Increase product usage by the existing customers. • Modify product. • Add new channels of distribution. Examples: (i) Classic movies are sold in videotapes. (ii) Arm & Hammer Baking Soda → (a) Baking agent, (b) antacid, (c) refrigerator deodorizer, (d) carpet deodorizer, (e) toothpaste. (iii) Selling cigarettes through vending machines and offering evening MBA programs. Marketers also sometimes try to extend product life cycle by adopting short term survival strategies such as pricing below the market and offering dramatic guarantees. However, these tactics rarely work in the long run as competitors can copy them easily. Some topics related to product life cycles: 1. Fashion, fad, and hype curve: When we discussed the product life cycle, we addressed the case of regular products which replace an inferior substitute and are eventually replaced by superior alternatives. There are three exceptions to the usual product life cycle called fashion, fad, and hype curve (saddle). Fashion: A fashion is a style (e.g., that of apparel, automobiles) that is popularly accepted and purchased by people over a reasonably long period of time. After that, it slowly dies out only to reappear after some time. Fashions (i.e., people’s preference for a given style) change constantly but follow a cyclic pattern. We observe a fashion cycle in product categories where a customer wants something different, but not too different from others. For example, for cars, there is a cycle as the shape of the car oscillates between square and rounded. Fad: A fad is a product with a novelty feature which enjoys meteoric success only to die out quickly. Unlike fashions, fads have a one time life, and they do not follow a gradual life cycle like regular products. Hype Curve: The hype curve, also called a saddle, occurs when the demand for the product first drops, but then the product gets a second life and a higher peak sales compared to the 101

first time. This may happen if there are two groups of customers, one group sensitive to indirect network externality (availability of other products to use with this product), and the other group is not. The insensitive customers buy immediately, giving the first peak. The sensitive customers wait for the contingent products to arrive and produce the second peak. 2. Planned Obsolescence → Strategy of making your own product obsolete by replacing it with another product marketed by you. Types of planned obsolescence: 1. Technological obsolescence (also called Functional Obsolescence): Replace your old product by a superior product. This is very common in high technology industries. It helps establish the reputation of an organization as a leader in research and development and allows it to charge a premium price. For example, Intel Pentium and Hewlett Packard LaserJet 4 were vastly superior to the previous generation of products offered by the same companies. These products allowed Intel and HP to move far ahead of the “clones.” Other organizations have also successfully adopted this strategy. Sun (manufacturer of computer work stations) follows a principle of doubling the computing speed of its products every year. 2. Fashion Obsolescence: Make superficial changes to make the old product out of fashion. This is used extensively by fashion designers, and to some extent by text book publishers. At one time, American car manufacturers also followed that strategy; competition from foreign cars has made that impossible now. When the strategy of planned obsolescence is criticized, the target is usually fashion obsolescence since it does not provide any tangible benefit to the customer.

6.5

Product features that affect rate of new product adoption

The following are the major factors that determine whether a new product will be adopted at all, and the speed of adoption if it is adopted: 1. Relative advantage This has several important implications: • If the advantage is in superiority in some aspect of functionality, then the marketer may position the product as different from product class, or even replacement for product class. Usually the product is offered at a premium price. This corresponds to the Point of Difference (POD) strategy. Example: Claritin. • If the advantage is in lower price, the similarity to product category is stressed. This corresponds to the Point of Parity strategy. • By making an itemized list of advantages and disadvantages relative to existing products and asking customers how much value they assign to each item, the marketer can set 102

an appropriate price for the new product. This is particularly important for a radically new product. (Even when a product is radically new, something else is attempting to fulfill the same customer needs and has a price. That price is the starting point.) 2. Compatibility There can be different dimensions of compatibility. Examples include: • Compatibility with beliefs (religious, political) • Compatibility with way of life. This can be very basic. A keypad where letters are arranged alphabetically was not accepted because consumers were not used to it. The metric system was never accepted in the USA. • Compatibility with products already owned. For example, a new software will be adopted more quickly if it can be used with existing hardware. 3. Complexity: Clearly, customers are less likely to adopt a more complex product. To counter this problem, the marketer has to educate the customer to use the product, and provide after sales service. 4. Trialability: Can you try out the product in small quantities? Can you lease the product for a limited time? 5. Observability: Is it easy to see the benefits of using the product? Is it easy to communicate the benefits of using the product?

6.6

Strategic Planning and Boston Consulting Group (BCG) Market Growth/Market Share Matrix

6.6.1 Introduction Many organizations market a wide range of products which are not always closely related. For example, the Altria group markets products ranging from cheese to tobacco. This phenomenon has occurred due to large scale mergers and acquisitions of the 1980’s which took place based on financial rather than marketing considerations. To manage a broad array of products, it is useful to group your products into “strategic business units” or SBU-s. A strategic business unit represents a grouping of one ore more related products or businesses within a multi-product firm with specific managers, resources, objectives and competitors. It is formally defined as a unit of the company that has a separate mission and objectives and that can be planned independently from other company businesses. An SBU can be a company division, a product line within a division, or sometimes a single product or brand. Dividing products into SBU-s provides better knowledge of the market and flexibility to act when presented with threats or opportunities. Thus, any organization can be thought of 103

as a portfolio of SBU-s. A given SBU may be a single product or a group of closely related products in the same market. A popular approach to the management of SBU-s is the “market growth/market share matrix” introduced by the Boston Consulting Group (commonly known as BCG) in the 1970-s. This method categorizes SBU-s according to two criteria: (1) The rate at which the market is growing. Typically, if this rate exceeds 10% per year, then it is considered high. Otherwise it is called low. In my class, I will use the 10% cut-off (over 10% is high, 10% or below is low.) (2) The relative market share of the SBU in that market. This expresses the market share of your SBU as a multiple of the market share of the largest competitor in the market. For example, if your SBU has a 20% market share and the largest 20% competitor has a 50% market share, then your relative market share is = 0.4. 50% In contrast, if your SBU has the largest market share of 20% and your largest 20% competitor has a market share of 10%, then your relative market share is = 2. 10% Relative market share is considered high if it exceeds 1 (that is, you have a larger market share than any competitor), and is considered low otherwise. Based on this approach, SBU-s of a firm are placed in four categories: 1. Star: High market growth rate, high relative market share. 2. Question Mark: High market growth rate, low relative market share. 3. Cash Cow: Low market growth rate, high relative market share. 4. Dog: Low market growth rate, low relative market share. The BCG recommends that firms should strive to achieve a portfolio consisting of both stars (for success and even survival in the long run) and cash cows (for cash generation at present). Dogs should be discarded. Regarding a “question mark,” the BCG recommends careful evaluation regarding whether it can be converted into a star. If yes, then money generated by cash cows should be used for that purpose; otherwise, it should be discarded. 6.6.2 Logic of the BCG Approach The BCG approach selects the two dimensions of market growth rate and relative market share for the following reasons: Market Growth Rate: It is desirable to operate in a high growth market to ensure the future success of the organization. In general, it is risky to rely only on SBU-s in low growth markets because of the existence of the product life cycle phenomenon that product categories pass through four distinct stages: introduction/market development, growth, maturity, and 104

decline. Low growth markets generally occur for mature product categories which only grow at the rate of the population growth. While many mature product categories have existed for a long time (e.g., vehicles with internal combustion engines, audio speakers, washing machines), demand can decline sharply if a superior alternative is introduced to the market. For example, record players were replaced rapidly by compact disc players. Operating in a high growth market has two other advantages. First, during growth, there is usually ample scope for product differentiation, and head to head competition can be avoided. Also, as the market is growing, sales increase even when market share remains the same. However, operating in a high growth market is often costly as an organization requires resources to maintain or increase its market share. Typically, a market grows fast when new customers start using a product, or new uses are found for an existing product. Either way, marketers need to generate primary demand for the product category by educating potential customers, which is an expensive proposition. Thus, a firm usually enters high growth markets for the sake of future and not present profits. Relative Market Share: In contrast to the desirability of high market growth rate which comes from product life cycle considerations, the desirability of high relative market share arises from two considerations: cost, and network externality. Cost: This is driven by two factors: (1) Economy of Scale: Economy of scale represents the fact that if you sell a larger quantity of a product, the cost/unit usually drops. The simple accounting explanation is that a fixed overhead is spread across a larger number of units produced, thereby reducing the average cost of a unit produced. However, economy of scale may happen due to many other factors. Some of them are as follows. First, if you are producing larger quantities of a product, the process of production is usually more efficient, which reduces the average cost of producing a unit of the product. Second, as you now need larger quantities of raw materials, you get better prices from your suppliers. Finally, you can distribute the product at a lower cost when you ship out larger quantities. For example, railroads charge you less per unit shipped if you make shipments in “carload amounts” (c.l.) rather than in “less than carload” (l.c.l.) amounts. While economy of scale effects are observed widely, there are cases where you may not find it. An example is when to increase production, you need scarce resources, which may include skilled labor. Then, a higher production rate leads to higher prices of such resources, eliminating the economy of scale. Another example is when more production means you have to pay overtime wages to the labor force, which is usually more expensive than regular wages. (2) Experience Curve: This concept was popularized in the 1970’s by the BCG, and is based on the proposition, supported by data from several industries (e.g., the electronics industry), that the cost of producing a unit of a product drops by a fixed percentage every time the cumulative volume of production doubles. For example, suppose the cost of producing a unit of a computer chip drops by 10 % each time cumulative production doubles, and the cost is $100/unit after 1000 units are produced. 105

Then, the cost/unit will be $90 after 2000 units are produced, $81 (i.e., 90% of $90) after 4000 units are produced, $72.90 (i.e., 90% of $81) after 8000 units are produced, etc. While the experience curve phenomenon is a very important finding in business management, we should be careful when we use it. Clearly, as you produce more units of a product, you will identify sources of inefficiency in production and find ways to eliminate them. However, there are two reasons I feel the approach is somewhat simplistic. First, the experience curve approach implies that cost goes down indefinitely as cumulative volume increases. That would not happen if the cost of the raw materials used is significant. However, if you consider only the production cost, it is more likely to occur. Second, the drop in cost of producing a unit of a product should depend on “research and development” (R & D) expenditure, which the BCG model does not consider. Note that economy of scale deals with the rate of production, while experience curve phenomenon deals with cumulative production. To summarize, note that if an SBU has a relative market share greater than 1 (i.e., the largest market share), it has a cost advantage over its competition due to economy of scale. This advantage grows with time due to the experience curve phenomenon as the company selling more also has higher cumulative production. An SBU with a high relative market share therefore has a sustained cost advantage over the competition which it can use to either increase its profits, or to squeeze out marginal competitors by reducing price. Network Externality: In addition to cost considerations, high relative market share is sometimes desirable due to the presence of a phenomenon called network externality. Network externality was first observed in network products such as telephone service and the internet, and it takes two forms: (1) Direct Network Externality: Sometimes, the value of a brand increases if more people use the brand. For example, the value of the word processor Microsoft Word is greater because it is used by more people. Thus, if you write a report on Word, you are fairly certain others can read it also. Note that if a brand has greater market share, it has greater value due to direct network externality. (2) Indirect Network Externality: This refers to the availability of contingent products that can be used with your brand. There are many products which are of value only if used with other (contingent) products. For example, the value of a personal computer derives from the software that can be used with it, and the value of a video game system (e.g., Playstation 3) depends on the number of game titles developed for it. Different machines in a product category often use different and incompatible formats (e.g., video game systems offered by Nintendo and Sony). The developers of contingent products are naturally attracted to formats with larger market shares as they represent greater demand potential. Thus, the formats with larger market shares gain value through greater availability of contingent products, and formats with smaller market shares often fail. Such failures include the technologically superior Beta format video cassette recorders which were eliminated by the VHS format machines. 106

Overall BCG Recommendations: The BCG approach recommends that an organization must balance the cash flow among its SBU’s and eliminate entities which do not help the process. Specifically: (1) Dogs should be discarded. (2) Stars are needed for long term success of the organization. However, stars require a large promotional effort to retain market share. This should be done by using the cash generated by cash cows. (3) Question marks should be evaluated closely. If they can be made into stars, then the cash generated by cash cows should be used to do that. Otherwise, they should be discarded.

6.6.3 Limitations of the BCG Approach The BCG approach remains popular, partly due to the simplicity of its ideas. However, it has some limitations: 1. It is not clear how a market is to be defined. Consequently, managers have been known to define their markets in a such a way that the relative market share looks high, thus making a case against the elimination of her/his brand. 2. The model ignores the possibility of the existence of synergy within a product line. A firm may find it necessary to keep a “dog” even when it loses money on it. For example, beer manufacturers often hold on to low- alcohol products simply to prove that they are socially responsible marketer, thereby defending itself against possible attacks on the other products they offer.

6.7

Appendix: Practice Problem on BCG Market Growth/Market Share Matrix

Acme Company has four strategic business units: (1) lap top computers, (2) color printers, (3) fax machines, and (4) digital cameras. The markets of the four strategic business units are as follows: (1) Lap top computer market: The market is growing 20% annually. Acme has a market share of 30%. The market shares of its competitors are: Dell: 20%, HP: 15%, IBM: 15%. The other competitors all have market shares less than 5%. (2) Color printers: The market is growing 20% annually. Acme has a market share of 18%. The market shares of its competitors are: HP: 30%, Canon: 20%, Epson: 10%. Other competitors all have market shares less than 10%. 107

(3) Fax machines: The market is growing 2% annually. Acme has a market share of 15%. The market shares of its competitors are: Sharp: 20%, HP: 15%. Other competitors all have market shares less than 10%. (4) Digital cameras: The market is growing 25% annually. Acme has a 15% market share. The market shares of its competitors are: Olympus: 12%, Canon: 10%, Kodak: 10%, HP: 10%, Microsoft: 8%. Other competitors all have market shares less than 5%. Questions: (i) Using the BCG growth/share matrix approach, how would you categorize the four strategic business units of Acme Company? (Show details of work) (ii) Using the BCG approach, what action should Acme take for each SBU? Describe the logic behind these recommendations. (iii) Consider the portfolio Acme would get if it follows BCG recommendations. Is this portfolio satisfactory? Clearly state yes or no and briefly discuss why you say so. If no, what type of SBU should Acme try to add to its portfolio to balance it?

6.8

Appendix: Hypothetical Example of Experience Curve Effect

According to the experience curve effect, the marginal cost of producing one more unit of the product drops by the same percentage whenever cumulative production doubles. Formally, the marginal cost C of producing another unit of the product after a cumulative production of Q units can be expressed as: C

=

C0 ∗ (

Q −η ) , Q0

where η is a fixed positive number, and C0 is the marginal cost after a cumulative production of Q0 units. A larger η represents a sharper experience curve. Examples of how fast marginal cost drops for different values of η are shown below, beginning with a marginal cost of $100 after a cumulative production of 1000 units. Cumulative Production 1000 2000 4000 8000 16,000

Marginal Cost if η = 0.152 $100.00 $90.00 $81.00 $72.90 $65.61

Marginal Cost if η = 0.235 $100.00 $85.00 $72.25 $61.41 $52.20

108

Marginal Cost if η = 0.322 $100.00 $80.00 $64.00 $51.20 $40.96

7

Selected Topics on Pricing Strategy

7.1

Introduction and Notations

In this section, we discuss the basic ideas and terms related to pricing. Here, we ignore any price modifications such as discounts, and focus on the basic price of the product. We also make simplifying assumptions about the costs of marketing. 7.1.1 Notations: 1. P = price/unit 2. Q = number of units sold 3. Revenue = P ∗ Q. This is the dollar sales. 4. Total Cost is given by: (7.1) Total Cost

=

Total Fixed Cost + Total Variable Cost = F C + (V C ∗ Q),

where: • F C = total fixed cost (does not depend on Q) • V C = unit variable cost. 5. Unit gross margin = P − V C. 6. Contribution to profit = (P − V C) ∗ Q. 7. Profit Volume Ratio (P V ratio, or simply P V ) =

P −VC . P

Note that Contribution to Profit = (P − V C) ∗ Q =

P −VC ∗ (P ∗ Q) = P V ratio ∗ Revenue. P

Thus, the P V ratio represents the fraction of revenue that becomes contribution to profit. 1 , gives the dollar sales required to generate one dollar of The inverse of the P V ratio, PV contribution to profit. 8. Profit, denoted by Π, is given by: (6.2) Profit (Π) =

=

Revenue − Total Cost

(P − V C) ∗ Q − F C

=

=

P ∗ Q − {F C + (V C ∗ Q)}

Contribution to Profit − Total Fixed Cost.

Example: Suppose you are a retailer selling shirts. You buy shirts at $10/unit from a manufacturer, and you sell them for $15/unit to the customer. You buy only as many shirts 109

as you sell. You pay $2000 as rent and utilities for your store. You have no other costs. Currently you sell 1000 shirts. In this case: • P = 15, Q = 1000, F C = 2000, and V C = 10. • Revenue is given by P ∗ Q = 15 × 1000 = $15,000. • Unit gross margin is given by P − V C = 15 − 10 = $5. 15 − 10 5 1 1 = = . Thus, for every dollar of sales revenue, dollars 15 15 3 3 1 become contribution to profit. Also, = 3, that is you need $3 of revenue to get $1 PV of contribution to profit.

• P V ratio =

• Total Cost = F C + (V C ∗ Q) = 2000 + (10 × 1000) = $12,000. • Contribution to Profit = (P − V C) ∗ Q = (15 − 10) ∗ 1000 = $5000. • Profit = (P − V C) ∗ Q − F C = (15 − 10) × 1000 − 2000 = $3000. 7.1.2 Break-Even Analysis: Assume that price exceeds unit variable cost (P > V C). If Q = 0, then profit = −F C. As Q increases from 0, profit increases from −F C. When the contribution to profit is equal to F C, we achieve the “break-even point” where the profit is zero, and the fixed cost has been exactly recovered. If sales exceed the break-even point, additional sales generate positive profits at the rate of P V dollars for each dollar of additional sales revenue. We call the number of units sold at the break-even point the Break Even Quantity (BEQ). This is given by: FC BEQ = . (P − V C) If Q exceeds BEQ, you have a positive profit. If Q is below the BEQ, you lose money. If Q = BEQ, there is neither profit nor loss: you have broken even. The break-even point in FC dollars, that is, the revenue that you need to break-even is given by . PV Example: Suppose you sell shirts, your F C = 2000, V C = 10, and P = 15. In this case, BEQ

=

FC P −VC

=

2000 15 − 10

=

400,

that is, you need to sell 400 shirts to achieve break-even. 7.1.3 Note: 1. Usually, the total fixed cost represents expenditure on machines needed for production, advertising, etc, and the variable cost represents the expenditure on labor, raw material purchase, supplies, etc. However, depending on the specific situation, these can change. 110

In a heavily unionized industry, labor costs may become fixed, regardless of the level of production or sales. Similarly, as demand increases, a firm may need to rent additional equipment to increase production, that is, equipment costs may change if production exceeds current capacity. 2. Unless otherwise stated, we assume that the unit variable cost (V C) is constant for all values of Q. That assumption is not always correct. For example, if the marketer receives quantity discounts on the purchase of raw materials, V C is lower if Q is higher. The same happens (V C lower if Q higher) if large-scale production corresponds the use of more efficient technology. On the other hand, if large Qrequires second-shift production and overtime wages, V C may be larger when Q is large.

7.2

Demand and Own Price Elasticity of Demand

7.2.1 Primary and Selective Demand: The demand for a product arises from the need customers feel for a product. A customer makes the following decisions: (1) Whether to buy a product from the product category at all? For example, should she purchase an MP3 player? If the product is a necessity of life, a purchase is likely to occur. (2) If a purchase is to be made, which one of the product alternatives to purchase? If the customer feels that many products fulfill similar needs, she is less likely to stay with the same product regardless of price. (3) How many product units to purchase? The demand for a product category (e.g., MP3 player) is called primary demand for the product. The demand for a specific alternative within the product category (e.g., Apple iPOD) is called selective demand for that product alternative. 7.2.2 Normal and Inverse Demand Curves: When a customer purchases a unit of a product, she gains the product unit in exchange for the price of the unit. The gain from the product unit depends on the attributes of the product. Clearly, the customer buys the product only if the gain from owning the product unit exceeds the price of the unit. Clearly, the demand for a product depends on the price of the product as well as its attribute values as perceived by the consumer. For a given product, we denote its demand in number of units by Q(P ) if the unit price is P . Demand can be either normal or inverse. Normal Demand: Usually, if P increases, Q(P ) decreases. This is called normal demand. For a normal demand curve, the plot of Q(P ) against P is a downward sloping line like that shown in Figure 7.1. 111

Q(P ) 6

-

P Figure 7.1: Normal Demand Inverse Demand: If a customer cannot evaluate product attributes directly, she often uses price as an indicator of product quality. Then, if the price is set too low, she may assume that product quality is unacceptably low and decide not buy the product. When many customers behave this way, we have inverse demand where the demand is lower if price is lower and higher if price is higher. Inverse demand may be observed experimentally by offering the same product at different price levels to different market areas. Inverse demand is more likely to happen for products like perfumes, fashion apparel, and jewelry, where product attributes are difficult to judge. Inverse demand does not occur if customers are sophisticated enough to evaluate product quality. Even when inverse demand occurs, it is only present when price is low. When price is high, we have normal demand once again (see Figure 7.2). Experienced marketers do not encounter inverse demand because they set price at a level the customer considers appropriate for the product offering. Because of that, in the discussions to follow, we assume that demand is normal unless stated otherwise. Q(P ) 6

Inverse ¾ Demand

- Normal

Demand

-

P Figure 7.2 112

7.2.3 Own Price Elasticity of Demand: This term captures how the demand (in units) of a product depends on P , the unit price of the product and is defined by: (7.3) ² = Price Elasticity of Demand =

% change in demand , % change in product’s price

where: • % change in price =

New Price − Old Price × 100%. Old Price

• % change in demand =

New Demand − Old Demand × 100%. Old Demand

P dQ d ln Q = ) Q dP d ln P Example: Suppose you are a retailer selling shirts. If price/unit is $20, the demand is 1000 units. If the price is changed to $21, the demand becomes 800 units. (More formally, using calculus notations: ² =

(21−20) × 100% 20 demand = (800−1000) 1000

In this case, the % change in price = The corresponding % change in

= 5%.

× 100% = −20%. − 20% Therefore, the price elasticity of demand, ² = = −4. 5% Note: (1) For a normal demand curve, the price elasticity of demand is negative, that is, an increase in price leads to lower demand. (2) Short Run versus Long Run Demands: Usually, when we talk about price elasticity of demand, we mean short run price elasticity of demand which is computed based on the immediate change in demand in response to a price change. Sometimes, the short run elasticity differs from the long run elasticity. Typically this is a result of sunk investment. For example, when oil prices rose sharply in the early 70’s, American customers were stuck with the cars they had, and the consumption of gasoline dropped only slightly in the short run. However, in the long run, customers switched to more fuel efficient cars, and gasoline consumption dropped significantly. 7.2.4 Price Sensitivity of Demand: We now restrict our attention to normal demand, that is, demand decreases if price increases, and vice versa (² < 0). For normal demand, we call |²|, the absolute value of price elasticity, the price sensitivity of demand, and denote it by the symbol E. For example, if ² = −3, then E = 3. If E is greater, then demand falls more sharply as price increases. 7.2.5 Relationship between Price Sensitivity and Revenue: We again assume normal demand and that demand Q(P ) equals the number of units sold (Q), that is, there is no unsatisfied demand. We now discuss how revenue (P ∗ Q) is affected by small changes in price. Three cases are possible: 113

(1) If E < 1, a 1% increase in P leads to a less than 1% drop in Q. Similarly, a 1% drop in P leads to a less than 1% increase in Q. Hence, revenue (P ∗ Q) increases if P increases, and decreases if P decreases. (2) If E = 1, a 1% increase in P is matched by a 1% drop in Q, and a 1% drop in P is matched by a 1% rise in Q. Hence, the revenue (P ∗ Q) does not change if price varies slightly from the current level. (3) If E > 1, a 1% increase in P leads to a greater than 1% drop in Q, and 1% drop in P leads to a greater than 1% rise in Q. Hence, revenue decreases in P increases, and increases if P decreases. Inelastic Demand: If E ≤ 1 (cases 1 and 2), revenue either remains the same or increases if price increases. This is called inelastic demand. Elastic Demand: If E > 1 (case 3), revenue decreases if price increases, and increases if price decreases. This is called elastic demand. Practice Problem: Suppose P = 100 and Q = 1000. (a) Compute the revenue at the present price. (b) Find the revenue if P = 101 for each of the following values of E: (1) E = .5 (2) E = 1.0 (3) E = 2.0 (4) E = 4.0 7.2.6 Relationship between price sensitivity and profit: We again assume that demand is normal and that there is no unsatisfied demand (Q(P ) = Q). From Section 1.1, Profit (Π) is defined as: Π

=

(P − V C) ∗ Q − F C.

In the expression for profit, only the term (P − V C) ∗ Q, is affected by changes in P . In this term (contribution to profit), both the factors (P − V C) and Q are affected by price. Effect of price change on (P − V C): It can be easily shown that for every 1% increase in 1 %. P , the unit gross margin (P − V C) increases by PV [Example. Suppose P = 20, and V C = 10, i.e., (P − V C) = 10. In this case, P V = 20 − 10 1 = .5, i.e., = 2. Suppose now P has increased to 21, i.e., there is a 5% increase 20 PV 11 − 10 in P . (P − V C) has increased to 11, i.e., there is a × 100%, i.e., 10% increase in 10 (P − V C). This is 2 times 5%, the percent increase in P .] Effect of price change on Q: From the definition of price sensitivity, for every 1% increase in P , the demand Q decreases by E %. 114

Combining, if P changes, the change in profit depends on which percentage change is 1 greater, , or E. Three cases are possible: PV Case 1. E < reduces profit.

1 PV

: In this case, an increase in price increases profit and a decrease in price

Explanation: Note that a 1% increase in price leads to a P1V % increase in (P − V C) and an E% drop in Q. As the % increase in (P − V C) here is more than the % drop in Q, the term (P − V C) ∗ Q and hence profit increases. Case 2. E =

1 PV

: In this case, profit does not change if price changes by a small amount.

Case 3. E > P1V : In this case, a price increase results in lower profit, and a price reduction leads to greater profit.

P −VC

Q

6

6

¢ ¢

¢

¢

¢

t¢

¢

t

¢

¢

¢

¢

¢

-

-

P

(P − V C)Q

(P − V C)Q

6

6

E<

P

1 PV

E>

1 PV

t

t

-

-

P 115

P

(P − V C)Q 6

E= t

1 PV

-

P

Figure 7.3 Practice Problem: Suppose V C = 10, Q = 1000, and E = 3. If you want to maximize profit, what, if anything, should you do at the following levels of price? (1) P = 11 (2) P = 12 (3) P = 15 (4) P = 20 (5) P = 30 1 7.2.7 Profit Maximizing Price: From Section 7.2.6, you can see that unless E = , you PV 1 can change price to increase profit. For example, if E > , you can decrease price and raise PV 1 profit, and if E < , you can increase price and increase profit. It can be shown that profit PV is maximized if and only if: 1 (7.4) E = . PV Two cases are possible: P −VC 1 Case 1: E ≤ 1: Note that for any P > V C, P V = < 1, that is, > 1. Thus, P PV 1 if E ≤ 1, you always have E < and can raise price to increase profit. This can happen PV when an essential product is in short supply and can lead to price gouging. Case 2: E > 1: If E > 1, equation (7.4) can be rewritten as: (7.5) P ∗

=

VC ∗

E E−1 116

=

VC +

VC , E−1

where P ∗ is the profit-maximizing price. Thus, the profit maximizing price is given by unit variable cost plus a mark-up that depends on E. If E is greater, the mark-up is smaller. 3 For example, suppose V C = 10. If E = 3, the profit maximizing price is 10 × = 15 3−1 (a 50% mark-up on variable cost). On the other hand, if E = 5, the profit maximizing price 6 = 12 (a 20% mark-up on variable cost). is 10 × 6−1

7.3

Factors That Affect Price Sensitivity

A customer has to decide whether she should buy the product at all (primary demand) and if yes, which alternative she should choose (selective demand). A related decision is how many units to buy. If she has many alternatives she finds similar, she will clearly avoid the more costly ones. This helps explain how price sensitivity depends on the different factors we now identify. 1. Price Relative to Purchasing Power: Usually, the price sensitivity of demand depends on the level of the price itself. In fact, E remains the same for all values of P only for the constant elasticity demand function discussed in Section 7.12.1. The reason why price sensitivity depends on price level can be seen by considering the price relative to the purchasing power of the customer. For example, suppose Sony offered its Vaio notebook computer with 1 GB RAM and 100 GB hard drive to SU students for $100/unit. Clearly, almost all students can afford to buy the product at both $100 and $110, and a 10% price increase from $100 would have no effect on demand. On the other hand, if these same machines were sold for $1500 each, a 10% increase in price to $1650 would have a large impact on demand as many students who would buy the product at $1500 would switch to a competing product or buy nothing at all. Example of change in elasticity with price level: Consider the linear demand function: Q

=

20, 000 − 100 P.

We will show that the price elasticity of demand is different for P = 100 and P = 150. (1) At P = 100, Q = 20, 000 − (100 × 100) = 10, 000. If P is increased by 1%, i.e., set at $101, new Q = 20, 000 − (100 × 101) = 9900, that is, Q changes by −1%. Therefore, at P = 100, ² = −1. (ii) At P = 150, Q = 20, 000 − (100 × 150) = 5000. If P is increased by 1%, i.e., set at $151.5, new Q = 20, 000 − (100 × 151.5) = 4850, i.e., Q (4850 − 5000) changes by × 100% = −3%. Therefore, if P = 150, ² = −3. 5000 2. Whether the product is considered a necessity or a luxury item: By definition, a consumer does not have a substitute for an item she considers a necessity, e.g., heating fuel during winter, or a pain reliever when she has a toothache. Clearly, the demand for necessities will be relatively inelastic in price as the consumer cannot do without them, while she does 117

not really need the luxury items and can postpone or avoid the purchase of them. These arguments hold for the demand of product categories (primary demand). If you consider the demand for specific brands (selective demand), the effect of price on demand is less clear. A staple product like bread is usually a commodity, and the consumer is likely to switch from brand to brand without hesitation. Thus, while the demand for the product category, bread, may be inelastic in price, the demand for a specific brand (e.g., Wonder Bread) is likely to drop sharply if its marketer raises price unilaterally. In contrast, luxury products (e.g., perfume, chocolate, stereo equipment) often exhibit high degrees of product differentiation. Thus, once the consumer has decided to buy the product, she may be less sensitive to price changes of a specific brand. 3. Awareness and availability of substitutes: If the consumers have many alternatives which satisfy their needs, they are more likely to switch in response to a price increase. The consumer’s price sensitivity therefore increases with an increase in the number of substitutes and the consumers’ awareness of them. 3.(a) If a product has unique values/features, the consumers are less likely to find a close substitute. As a result, increased product differentiation reduces price sensitivity. Example: Decaffeinated coffee, lactose-free milk, organic vegetables. 3.(b) Price quality association usually comes with reduced price sensitivity. One possible reason is risk aversion. We discuss this factor in more detail in the next section. 3.(c) If the alternatives are more difficult to compare (e.g., different cold remedies), consumers are less price sensitive. Note that here, lower price sensitivity is very similar to higher brand loyalty. 4. The size of expenditure involved for consumer: 4.(a) Shared Cost Effect: If somebody else pays partly or fully for the product purchase, the consumer is likely to be less price sensitive. Example: Patients covered by comprehensive health insurance purchasing medicine, business travelers purchasing airline tickets. Marketer’s Perspective: Since customers are not price sensitive, it is difficult to attract customers by offering price cuts. Possible Strategy: Give non-price incentives to customers for trying out product. For example, a frequent flyer plan may attract the attention of business travelers. 4.(b) Heavy users are usually more price sensitive than light users of a product. Possible Reasons: (i) Greater knowledge of alternatives. (ii) More suppliers compete for the heavy users, thereby increasing the number of alternatives. 4.(c) If the product purchase is part of a larger purchase/expenditure, price sensitivity decreases. For example, if restaurants in New York City charged a higher price for food, the price increase would be more likely to reduce local demand than demand from people coming from out of town to spend the day in New York City. 118

5. Consumer’s ability to take advantage of a lower price. 5.(a) Inventory Effect: If the product can be stored for future use, price sensitivity is greater. For example, a person with a large house can take advantage of a sale price on breakfast cereals more easily than a student who lives in a dormitory. 5.(b) Sunk Investment Effect: Many purchases are used in conjunction with assets bought previously, e.g., toner cartridges with printers, refills with pens, etc. To the extent that the investments in the printers, pens, etc., are sunk, the price sensitivity for the products used with them is reduced. Example: When gas prices went up in the early 1970’s, consumers were stuck with large cars. In the short run, demand for gas dropped only slightly. However, in the long run, consumers purchased more fuel efficient cars, and demand for gas dropped sharply.

7.4

Price Quality Relationships

When a consumer cannot judge a product completely in terms of its attribute levels, she uses the price of the product as an indicator of quality. This is often the case for new products. Because of this phenomenon, a marketer has to be very careful about setting the price of a new product. For example, she may successfully introduce a product at a low price to get market share. However, this may create an image of low quality, and if she subsequently tries to raise the price, demand may drop sharply. Japanese auto makers have avoided this problem by continuing to modify and improve their product offerings. Marketers often create an impression of quality with higher prices. For example, faced with competition (from Wolfschmidt), Smirnoff Vodka actually raised its price to set it apart. Some reasons why this works: (i) Buyers justify higher expenditure by believing that quality is higher. (ii) Snob appeal: The purchase of a higher priced product allows the purchaser to separate himself from the masses. Example: High fashion, wine, perfume, jewelry. (iii) Risk aversion, as customers assume a low price indicates low quality and do not want to take a chance. (Example: Michelin ran an advertising campaign with theme “So much is riding on your tires” stressing that one’s child is safer with Michelin.)

7.5

Basic Methods of Setting Price

In practice, the marketer rarely knows the exact relationship between demand and price. This is the main reason why we have many different methods of setting the price of a product. 7.5.1 Price Determination in Economic Theory: This is the hypothetical ideal case when the marketer can exactly quantify how demand depends on price. Since the marketer produces and sells the exact number of units the consumers demand, Q, the quantity sold, is 119

equal to demand. Standard economic theory assumes that the marketer sets price to maximize profit. This is achieved when the marginal cost of making another unit of product is equal to marginal revenue (incremental revenue from selling that additional unit of product). The profit maximizing price is given by equation (7.4). 7.5.2 A Simple Version of Economic Theory: In practice, it is virtually impossible for a marketer to know the exact relationship between Q and P . Still, a sophisticated marketer usually knows price sensitivity (E) around the current level of price. Using this knowledge, she can apply the results of Section 2.6 to make small adjustments in price to improve profitability: • If E <

1 , she can increase profit by increasing price. PV

• If E >

1 , she can increase profit by reducing price. PV

• If E =

1 , she should not change price; profit is already maximized. PV

Example: Suppose you are selling shirts at $15/unit, and the variable cost is $10/unit. 1 = , i.e., P1V = 3. Here, P V = (15−10) 15 3 In this case, unless E ≥ 3, an increase in price increases profit. Important Note: The following results hold for the profit maximizing price: • The profit maximizing price maximizes contribution to profit also. • The profit maximizing price does not depend on fixed cost. 7.5.3 Cost Oriented Pricing: Usually a firm can reasonably estimate its costs, but cannot estimate demand with precision. In that case, a cost oriented method is commonly used. I give two examples of cost based pricing. (1) Fixed Percentage Mark-Up Method: Retailers use this method quite often. By definition, selling price, P = V C + Mark Up. In the fixed percentage mark-up method, the mark up is set equal to a pre-specified fraction of V C. Example: Suppose V C = 10. If we use a mark-up equal to 20% of the variable cost, the selling price P = 10 + 0.2 × 10 = 12. (2) Target Profit Method: In this method, the marketer decides up front how much profit it wants to make if it sells a given quantity of the product (number of units), Q. Let ΠT denote the target profit. Then, we must have, ΠT = (P − V C) ∗ Q − F C −→ (P − V C) ∗ Q = (ΠT + F C). 120

Rewriting, P

=

VC +

(ΠT + F C) . Q

Note that here, the mark-up is inversely related to Q. Example: Suppose you sell shirts. V C = 10, F C = 2000, and you want to make a profit of $1000 by selling 2500 shirts. (1000 + 2000) = 11.20. Using the target profit method, P = 10 + 2500 Advantages of Cost Oriented Pricing: 1) Usually, it is possible to estimate costs. 2) Helps determine what levels of sales are needed for profitability. 3) If price is predetermined, tells a firm what the costs should be. Disadvantages of Cost Oriented Pricing: 1) Assumes that the consumer demand is not be lower than the quantity the firm intends to sell. Does not actively incorporate consumer preferences, competitive action, etc. 2) Sometimes it is very difficult to measure the cost of producing a product. Some reasons: (i) The firm uses the same manufacturing facilities to produce multiple products, and the total costs have to be allocated among the products. (ii) Costs change with experience and/or scale of production. (iii) The costs of labor, raw materials, etc., fluctuate rapidly. 7.5.4 Competition Oriented Pricing: 1) Pricing to meet competition (Status Quo Pricing). This strategy is appropriate when customers are very price sensitive, a situation often observed in mature product categories. When a competitor changes price, a firm usually follows that price change. In practice, this method is often used when the firm does not know the exact price sensitivity of demand but feels that it is large. This is often the case when salespeople have the authority to modify price. Product differentiation reduces the need to match price changes by the competition. Advantage: This is a safe strategy. 2) Pricing above the competitive level. This is often associated with a prestige objective. 3. Pricing below the competitive level. Typically, this is the pricing strategy of a generic product. 7.5.4 Psychological Pricing: This is a class of pricing techniques which includes odd-even pricing and customary pricing. I discus it separately in section 7.7 of the Reader. 7.5.5 Value-Based Pricing: Here, the price is based on the product’s perceived value to the consumer, including both benefits and risks. The value is based on the features of the augmented product rather than the core product, and a firm can use the non price variables in its marketing mix (e.g., advertising, packaging) to build up perceived value in the buyers’ minds. To apply this method, the marketer begins with the price of an alternative product as a reference point, and makes an itemized list of advantages and disadvantages of his own offering 121

compared to this alternative. Survey research is used to determine how much monetary value the customer places on each item. Adding these price modifications to the price of the reference product, the marketer determines what price he should charge. Example.2 Caterpillar Tractors: Price of tractor is $24,000. Price of competitor’s tractor is $20,000. (These were the prices in the late 1970’s.) Justification: $20,000 = Competitor’s Price. $ 3,000 = Price premium for superior durability. $ 2,000 = Price premium for superior reliability. $ 2,000 = Price premium for superior service. $ 1,000 = Price premium for longer warranty on parts. $28,000 = total value of package relative to competition $ 4,000 = discount $24,000 = final price

7.6

Competitive Bidding

Basic Idea: Government agencies and non-profit organizations often ask the marketer to submit a sealed bid. The marketer wins the contract if his bid price is lower than the competitors’ bids. The product specifications and the quantity sold if the marketer wins the bid are predetermined. We now discuss one method of selecting an optimal bid price, the Expected Profit (EP) method. Expected profit is defined as follows: Expected Profit = P(Win at Bid Price)∗(Bid Price − Cost), where: • P(Win at Bid Price) = the probability of winning the bid at the bid price. • Cost = estimated total cost. Clearly, the probability of winning, increases if Bid Price is lower, but the profit if the marketer wins, (Bid Price − Cost) decreases if the Bid Price is lower. The objective of the marketer is to choose Bid Price to maximize Expected Profit, which is the product of P(Win at Bid Price) and (Bid Price − Cost). The probability of winning at different bid prices is determined from historical data. Example: Suppose a school district has invited bids from local vendors to supply 1000 desks. You are a vendor, and you estimate your cost of supplying the 1000 desks as specified to be $80,000. You estimate the following probabilities of winning the bid at different bid prices: 2

Source: Thomas T. Nagle, The Strategy and Tactics of pricing, edition 1, Prentice Hall, 1987

122

Bid Price $80,000 $90,000 $100,000 $110,000 $120,000 $130,000

Probability of Winning 1.00 0.80 0.60 0.30 0.10 0.00

Which of the bid prices listed above ($80,000, $90,000, $100,000, $110,000, $120,000, and $130,000) will you choose? Augmenting the table to add the profit if you win and the expected profit, we get: Bid Price

Probability of Winning

$80,000 $90,000 $100,000 $110,000 $120,000 $130,000

1.00 0.80 0.60 0.30 0.10 0.00

Profit if win (Bid Price − Cost) $0 $10,000 $20,000 $30,000 $40,000 $50,000

Expected Profit $0 $8000 $12,000 $9000 $4000 $0

Hence, we choose a bid price of $100,000 to maximize expected profit. Application and Refinement: Clearly, the key challenge in setting bid prices is to determine the probability of winning at a given bid price. To determine these probabilities, you need to get data on the product specifications and the winning bid prices from the past. Method 1: The simple but crude method is to proceed as follows: • For each bid in the data set, estimate your cost of providing the product as specified. Call it C. Compute the “winning bid ratio” R∗ = P ∗ /C, where P ∗ is the winning bid price. • Consider a given bid ratio R = P/C, where P is the bid price. For this R, from the historical data, find what percentage of cases where R∗ exceeds this R. This percentage is an estimate of the probability of winning at this bid ratio. For instance, in the example above, the probability of winning at a bid ratio 1.25 is 60%. Method 2: If you can obtain not just the winning bid price, but all bid prices, you can refine your estimation of probabilities. Proceed as follows: • For each bid in the data set, estimate your cost, C. Compute the bid ratios for all the bid prices. For instance, suppose the cost is $100,000, and the bid prices are $80,000, $100,000, and $150,000. Then, the bid ratios are 0.8, 1.0, and 1.5. 123

• For any specific bid ratio R, find the fraction of bid ratios that exceed R. Call this number P (R). This is the probability that any one given competitor will have a bid ratio exceeding R. • Suppose you have selected a bid ratio R, and there are N other bids made by your competitors. You will win if all competing bids are higher. Assuming independence, the probability of winning is P (win|R) = {P (R)}N . • Compute P (R) and P (win|R) for all bid ratios under consideration. Example: A government agency wishes to purchase 2000 laptop computers and has invited sealed bids. You are a laptop computer vendor, and you estimate that your total cost to supply the laptops as specified is $1,000,000. You are trying to choose one of the following five bid prices: (1) $1,100,000, (2) $1,200,000, (3) $1,300,000, (4) $1,400,000, and (5) $1,500,000. You have obtained 20 bid prices submitted by competing vendors in the past; the bid ratios for these 20 bids are as follows: Case 1 2 3 4 5

Bid Ratio 1.008 1.107 1.229 1.267 1.281

Case 6 7 8 9 10

Bid Ratio 0.703 1.376 1.284 1.293 1.080

Case 11 12 13 14 15

Bid Ratio 1.594 0.911 1.089 1.327 1.214

Case 16 17 18 19 20

Bid Ratio 1.346 1.456 1.580 1.326 1.344

Suppose you know that other than you, two competing vendors will submit bids. Which of the five bid prices will give you the highest expected profit? Solution: You are considering five bid ratios, 1.1, 1.2, 1.3, 1.4, and 1.5. From the historical data, you have the following results: Bid Ratio (R) Fraction of bids with bid ratios higher than R (P (R)) 15 1.1 = 0.75 20 1.2

14 = 0.70 20

1.3

8 = 0.40 20

1.4

3 = 0.15 20

1.5

2 = 0.10 20

For example, consider the bid ratio 1.2. Since you have two competing bids, you will win if both competing bids have higher bid ratios than 1.2. The probability of that happening is 124

P (1.2)2 = 0.72 = 0.49. Thus, we have the following results: Bid Price 1,100,000 1,200,000 1,300,000 1,400,000 1,500,000

Bid Ratio (R) P (R) 1.1 0.75 1.2 0.70 1.3 0.40 1.4 0.15 1.5 0.1

P (win) 0.5625 0.49 0.16 0.0225 0.01

Profit if Win 100,000 200,000 300,000 400,000 500,000

Expected Profit 56,250 98,000 48,000 9000 5000

Therefore, out of the five bid prices considered, the bid price of $1,200,000 maximizes expected profit.

7.7

Psychological Pricing

Consumers do not always behave as a traditional economist would believe. Here we discuss two instances of “irrational” behavior, and offer possible explanations. 1. Odd-Even Pricing: It is commonly believed that buyers perceive odd-number prices to be significantly lower than the slightly higher round numbers they approximate. Examples: $1.99 vs $2.00, $19.95 vs $20.00, $5995 vs $6000. Traditionally, odd prices have been used to indicate bargain values, and even prices to indicate quality. Two possible explanations for the present use of odd pricing (neither very satisfactory): (a) Compare the two pairs: (i) $89 vs $78. (ii) $90 vs $79. Which one is a better bargain? The idea: When a consumer compares two prices quickly, he compares the digits from the left to the right. For example, when comparing $89 vs $78, he will first compare 8 with 7. When comparing $90 vs $79, he will first compare 9 with 7. This way, the second pair will appear to represent the better bargain, even though the price difference is the same in both cases, and the difference is a greater percentage of the price for the first pair. (b) Odd pricing has been used for a long time. As a result, consumers have learnt to associate odd prices with bargains. Thus, odd pricing has become a customary pricing method. 2) Customary Pricing: Due to custom and tradition, consumers expect what the price should be for certain products, e.g., $1.00 for a candy bar at a vending machine, etc. In these cases, a perceptible change in the price may lead to a sharp drop in demand. Marketers counter this in different ways. One method commonly used for frequently purchased packaged goods is to keep price/package the same, but reduce the amount of product in a package. This practice has led consumer advocates to insist that supermarkets use unit pricing in which all prices are stated in terms of some recognized unit of measurement or a standard numerical count. Another method is to change the price in very small steps (e.g., price of postage stamps), so that the consumers do not notice the change. 125

7.8

Skim and Penetration Pricing

Both skim and penetration pricing strategies can be explained by equation (7.5). From equation (7.5), the profit maximizing price depends on both E and V C. If E is larger, the price is smaller. In contrast, if V C is larger, the price is larger. It can also be shown that if V C is larger, profit is smaller. Skim Pricing: Skim pricing is sequential price discrimination. The marketer divides the market into groups based on price sensitivity. First, the price is set to maximize profit from the least price sensitive customers (lowest E and hence highest P ∗ ). Once the potential of this segment is exhausted, price is set to maximize profit from the next least price sensitive customers. Some issues related to skim pricing: (1) Charging a high price attracts competitors to enter the market. Thus, skim pricing is appropriate for unique products with good patent protection. Examples include pharmaceutical products which are introduced at a high price (e.g., Cipro, Claritin, AZT). (2) Profit maximizing price depends on V C also. As a higher price limits your demand, you are giving up benefits from economy of scale and experience curve effects by using skim pricing. Thus, skim pricing is appropriate if such effects are not strong. (3) Skim pricing may be used during the introduction of a new product when you wish to maintain product quality by limiting demand. (4) If there is strong price quality association, skim pricing is appropriate as a high price induces quality-conscious customers to buy the product and also serves as a barrier to competitive entry. Penetration Pricing: From equation (7.5), the profit maximizing price depends on V C and is lower if V C is smaller. It can also be shown that if V C declines, profit (Π) is always higher. Penetration pricing is used when a marketer feels that by offering a lower price now, V C can drop fast from economy of scale and experience curve and thus generate more profits in the future. Issues related to penetration pricing: (1) Here, the objective is to maximize long term profit instead of short term profit. This is consistent with the BCG approach which focuses on having the largest market share. (2) Penetration pricing can create a barrier to competitive entry if a strong experience curve effect exists. (3) Penetration pricing is appropriate for functional products where the customers can easily assess product attributes. That way, customers do not automatically think that quality is lower just because price is lower. (4) You have adequate production capacity, and good quality control to ensure product quality at a large production volume. Examples: Texas Instruments used a penetration pricing strategy for its calculators. Japanese 126

automobile companies also used the penetration pricing strategy successfully.

7.9

Spatial (Geographic) Pricing

Note: This is quite different from “geographical pricing” discussed in the Kotler/Keller text. A marketer sometimes serves different customers with different variable costs. For example, suppose you are a pizza delivery store, and the variable cost of making a large pepperoni pizza is $5. However, customers are scattered over a large area, and the cost of delivering the product increases with distance from the store. In that case, your unit variable cost is the base unit variable cost ($5) plus the cost of transportation. Even if your customers have the same level of price sensitivity everywhere, equation (7.5) suggests that the optimal price should be different at different locations due to variations in the cost of transportation. The study of how prices should be modified due to spatial dispersion of customers is called geographic pricing or spatial pricing. While it is possible to use equation (7.5) to determine the profit maximizing prices for all customers, marketers use simple rules to set the prices for customers at different distances. Four commonly used rules are discussed below. In these discussions, the term delivered price is the unit price the customer pays after considering the cost of transportation. (1) Uniform Delivered Pricing Plan: Here, the marketer charges all customers the same delivered price. Note that the gross margin of the marketer decreases with increasing distance from the source. Thus, a marketer using uniform delivered pricing typically serves customers within a specified boundary. Also, customers located close to the marketer effectively subsidize customers at a greater distance. Examples: Pizza delivery services usually offer free delivery within a given boundary; the US Postal Services charges the same fee of 41 cents for a standard letter for any destination within the USA. (2) Free on Board Origin (FOB origin) Pricing Plan: Here, the marketer charges the customer a base price plus the exact cost of transportation. Thus, the unit gross margin is the same for all customers, and thus the marketer does not have to limit its market served. However, customers at a greater distance face a higher delivered price and thus may decide not to patronize the store. The FOB origin plan can be implemented simply by making the customer entirely responsible for transporting the product from its source. For example, all grocery stores and eat-in restaurants implement FOB origin pricing automatically. (3) Zone Pricing Plan: This is a direct extension of the uniform delivered plan where the marketer divides the market into zones and offered the same delivered price to each zones. By increasing the number of zones, the marketer can increase contribution. However, the fixed cost of implementation also increases with the number of zones used. As a result, marketers using zone plan typically use a small number of zones. (4) Freight Absorption Plan: Here, the marketer pays all or part of the transportation cost. 127

You can think of FOB origin pricing (no freight absorption) and uniform delivered pricing (total freight absorption) as two extreme cases of the freight absorption method.

7.10

Cross Price Elasticity of Demand and Pricing Related Products

7.10.1 Cross price elasticity of demand: We now want to quantify how demand changes if your price remains fixed, but the price of another product changes. Definition: Let ²AB denote the cross elasticity of the demand of product A with respect to the price of product B. Then: ²AB =

% change in demand of product A % change in price of product B

P ∂QA (In calculus notations: ²AB = B .) QA ∂PB

Example: Suppose you sell shirts, and the current demand for the shirts you sell is 2000 units. A competitor sells shirts at $20/unit. If you do not change price, but the competitor increases his price to $21/unit, the demand for the shirts you are sell becomes 2150. (21 − 20) × 100% = 5%. 20 (2150 − 2000) The resulting change in the demand for your product = × 100% = 7.5%. 2000 Therefore, in this case, the cross price elasticity of demand for your product, relative to the competitor’s price = (7.5%)/(5%) = 1.5. Here, change in the competing product’s price =

Substitute products and Complementary products: (1) If two products have positive cross elasticities of demand relative to each other’s prices, they are substitute products, that is, a higher price of one product leads to switching to the alternative product. Clearly, for substitute products, the cross elasticity of demand decreases with increase in product differentiation. (2) If product A has a negative cross price elasticity of demand relative to product B, then A is a complement to B. In this case, a higher price for B leads to lower sales for B, which, in turn, reduces sales for A. Example: Toner cartridges are complementary products to printers. Note on computing cross price elasticity of demand: Demand at new price − Demand at old price × 100%. % change in demand = Demand at old price New price − Old price % change in price = × 100%. Old price 7.10.2 Pricing related Products: So far, we discussed the pricing of a single product. Here, instead of one product, we consider the simultaneous pricing of two or more related products. 128

Economic Issues (based on cross elasticities): 1) Pricing Substitute Products: Suppose you sell two substitute products A and B (e.g., regular unleaded and premium unleaded gasoline), and the price of product A is set at the profit maximizing level given by equation (7.4). Suppose now you increased the price of product A by 1%. As discussed in Section 7.2.6, the demand of product A will decrease by 1 E%, the gross margin/unit of product A will increase by %, and on balance contribution to PV profit from product A will not change. However, some customers who no longer buy product A will buy the substitute, product B. Thus, the contribution to profit for product B will increase. Combining, the total profit from products A and B will increase. To summarize, if you sell two substitute products, the profit maximizing price of each product is higher than the profit maximizing price if you consider the product in isolation. 2) Pricing Complementary Products: Suppose you are sell two complementary products A and B (e.g., printers and toner cartridges). Suppose the price of product A is set at the profit maximizing level given by equation (7.4). If you increase the price of A by 1%, the demand 1 for A drops by E%, the gross margin/unit increases by %, and on balance the profit from PV A does not change. However, the demand for B also drops, and your combined total profit from A and B drops. In contrast, if you reduce the price of A, the total profit from A and B combined increases. Thus, when you sell two complementary products, the profit maximizing price of each product is lower than the profit maximizing price of the product considered in isolation. Example 1: Gillette sells its razors at a low price in order to attract customers who will subsequently buy the razor blades (note the sunk investment effect here). Example 2: Loss Leader Pricing. Sometimes a marketer sets the price of a product very low (sometimes even lower than its variable cost) to attract customers to the other products it offers. Supermarkets regularly take losses on advertised items (e.g., white bread for 39 cents, etc) to attract customers to the store. There are legal restrictions on the use of this strategy. A retailer can sell a given product at or below cost for only a limited amount of time. However, he can still offer “special prices” for different products during different time periods.

7.11

Additional Topics

7.11.1 Price Discrimination: Contrary to common belief, a marketer can legally discriminate among ultimate consumers based on price. In fact, price based segmentation is another name for price discrimination. Basic Idea: Different segments have different levels of price sensitivity. Thus, the optimal price will differ from one segment to another. Price discrimination is simply an attempt to charge the less price sensitive customers a higher price and offer a lower price to a more price sensitive customer. 129

Examples of Price Discrimination Basis of discrimination Examples Income, earning power Doctor’s fees. Professional association dues. Tuition rate. Age, gender, student status Hair-cut. Concert tickets. Senior citizen fares. Location of buyer Grocery prices differ between inner city and suburbs. Status of buyer New subscriber to a magazine. Size of purchase Package size, number of units ordered. Usage rate Long distance telephone rates. Time of purchase Season/off-season, peak/off-peak. Note: Price discrimination works well if “leakage” is not significant, i.e., if consumers who get a lower price cannot undercut the prices of the original seller, and capture the price-insensitive market. This is why we find widespread use of price discrimination in the marketing of services where the service itself cannot often be separated from the service provider, thereby eliminating the possibility of resale. A doctor can therefore charge different fees to patients with or without an insurance coverage, and a bus line can give a lower fare to senior citizens. 7.11.2 Quantity Discount: This may be offered to resellers as well as final consumers. Some reasons why a quantity discount may be offered: 1) Segmentation based on price sensitivity: Heavier users may be more price sensitive and a lower price is therefore appropriate for them. Long distance telephone companies give lower rates to heavier users. (This is an example of a cumulative quantity discount.) 2) Order discount: Sometimes, the cost for preparing an order is about the same for all order sizes. In that case, the seller gives discounts for making larger orders. Objective: Induce buyer to make one large order rather than several small ones. (This is an example of non-cumulative quantity discount.) 3) Increase demand, which, in turn, reduces the cost of production by exploiting the economy of scale, usage of inactive plant capacity, etc. This can also help clear inventory. 4) Coordination in a channel of distribution: A manufacturer may use a quantity discount to affect the pricing strategy of a retailer and thereby coordinate activities of different levels of the channel of distribution. 7.11.3 Robinson-Patman Act: Excerpts: “It shall be unlawful for any person engaged in commerce . . . to discriminate in price between purchasers of commodities of like grade and quality − where either or any of the purchasers involved in such discrimination are in commerce, − where the effect of such discrimination may be substantially to lessen competition or tend to create a monopoly in any line of commerce, or to injure, destroy, or prevent competition 130

... Important Points: 1. The objective of the Robinson-Patman Act is to maximize competition and prevent formation of monopolies. For example, if a manufacturer sells two competing retailers the same commodity at different prices, that hurts competition and is illegal. However, since a wholesaler and a retailer usually do not compete directly, the manufacturer can offer different prices to them. 2. The Robinson-Patman Act only applies when the competing buyers are business customers. 7.11.4 Buy-Response Curve In general, it is very difficult to determine demand at every level of price. One method which is crude but easy to implement is called the buy-response curve. This method uses a survey of potential customers. A respondent is asked to consider a specific product and tell how likely she is to purchase the product at different price levels. By aggregating data over a sample of respondents, the researcher gets a crude idea of demand at every price level. The data can also be fitted using the regression model to obtain an estimate of the price elasticity of demand. Examples of Buy-Response Curves Example 1. Survey Question: Suppose the band Rolling Stones plans to offer a concert at the Carrier Dome. Assuming that you have the time to attend the concert, please indicate how likely would you be to buy a ticket for this concert at each of the following ticket prices by circling your response. (1 → definitely not buy, 4 → 50/50 chance, 7 → definitely buy.) Price $40 $45 $50 $55 $60 $65 $70 $75 $80

Definitely Not Buy 1 1 1 1 1 1 1 1 1

2 2 2 2 2 2 2 2 2

3 3 3 3 3 3 3 3 3

4 4 4 4 4 4 4 4 4

5 5 5 5 5 5 5 5 5

6 6 6 6 6 6 6 6 6

Definitely Buy 7 7 7 7 7 7 7 7 7

Data were collected from a sample of 44 MBA students. For each item, I recorded 1 if 5 or more was circled, and 0 if not. Then, I counted how many respondents scored 1 and assumed that those students would purchase the product at that price level. For example, in the table below, 22 out of the 44 students will buy the ticket at a price of $40. The demands are as follows: 131

Price (P ) Demand (Q)

40 45 22 18

50 55 60 15 12 8

70 75 4 2

Clearly, we have a normal demand curve here. Example 2. Survey Question: How interested are you in buying a ticket for an exhibition basketball game between Denver Nuggets and Cleveland Cavaliers at the Carrier Dome at each of the prices listed below? Please circle.

$20 $25 $30 $35 $40 $45 $50 $55 $60

Not interested at all 1 1 1 1 1 1 1 1 1

2 2 2 2 2 2 2 2 2

3 3 3 3 3 3 3 3 3

4 4 4 4 4 4 4 4 4

5 5 5 5 5 5 5 5 5

6 6 6 6 6 6 6 6 6

Very interested 7 7 7 7 7 7 7 7 7

Data were collected from a sample of 71 Syracuse University undergraduate students. For each item, I recorded 1 if 5 or more was circled, and 0 if not. Then, I counted how many respondents scored 1 and assumed that those students would purchase the product at that price level. The demands for men, women, and the whole sample are as follows: Price (P ) Men (40) Demand (Qm ) Women (31) Demand (Qw ) All (71) Demand (Qall )

20

25

30

35

40

45

50

55

60

36 90.00%

36 90.00%

33 82.50%

29 72.50%

29 72.50%

27 67.50%

16 40.00%

13 32.50%

12 30.00%

21 67.74%

18 58.06%

16 51.61%

16 51.61%

12 38.71%

9 29.03%

3 9.68%

2 6.45%

1 3.23%

57 80.28%

54 76.06%

49 69.01%

45 63.38%

41 57.75%

36 50.70%

19 26.76%

15 21.13%

13 18.31%

Example 3. Survey Question: If a combined lacrosse/football/basketball season ticket were available, how likely would you be to purchase it at the following prices? (Please circle) 132

$50 $75 $100 $125 $150 $175 $200

Very unlikely 1 1 1 1 1 1 1

2 2 2 2 2 2 2

3 3 3 3 3 3 3

4 4 4 4 4 4 4

5 5 5 5 5 5 5

6 6 6 6 6 6 6

Very likely 7 7 7 7 7 7 7

Data were collected from 148 Syracuse University Undergraduate students. Demand is expressed as percentage of population that circled 5, 6, or 7 at a given price. P 50 75 100 125 150 175 200

7.12

Q (overall) 87.07% 87.07% 76.19% 61.9% 47.62% 36.55% 28.08%

Q (Women) 82.67% 81.33% 69.33% 57.34% 44% 28.38% 25.34%

Q (Men) 91.66% 93.06% 83.33% 66.68% 51.38% 45.07% 31%

Demand Functions used in Practice

7.12.1 Two Commonly Used Demand Functions: While the normal demand curve applies in typical pricing situations, there is no universal agreement about the functional form of the demand curve. Two demand curves, the constant elasticity (also called Cobb-Douglas) demand curve, and the linear demand curve, are used extensively in practice. These two demand curves are illustrated graphically in Figure 7.4. Q 6

Q 6 A

@ @

@

@ @

@

@ @

@

0 0

@ @

0

-

0

P Constant Elasticity Demand

Linear Demand Figure 7.4 133

@ -

A/B

P

1. Constant Elasticity (Cobb-Douglas) Demand Curve: This curve assumes the following relationship between demand and price and is named after two U.S. Congressmen who discovered it: (7.6) Q = KP −E , where K and E are positive constants. It can be shown that E is equal to price sensitivity for all levels of price. If this demand curve applies, then profit maximizing price can be determined by using equation (7.5) if E > 1. If E ≤ 1, a profit maximizing marketer will raise price as long as it is legally allowed. Example: Suppose the demand function is 1000 × P −3 , and V C = 10. Here, K = 1000 and 3 E = 3. Hence, the profit maximizing price is 10 × = 10 × 1.5 = 15. 3−1 Note that if V C increases by $1 to $11, P ∗ increases to 11 × 1.5 = 16.5, that is, P ∗ increases by $1.5. 2. Linear Demand Function: This curve assumes that demand can be expressed as follows: (7.7) Q = A − B ∗ P

if P ≤

A , B

and Q = 0 if P >

A , B

where A and B are positive constants. For the linear demand function, E depends on the level of P , and equation (7.5) cannot be used directly to find the profit maximizing price. For this demand curve, the profit maximizing price is given by: (7.8) P ∗

=

1 A ( + V C), 2 B

that is, the profit maximizing price for the linear demand function is the average of (1) the price point A/B where demand becomes zero, (2) the unit variable cost (V C) (see Figure 7.5). Q

6 @

@

@

@

@

@

@

@

@

@

@

@

@

@

VC

P∗

-

@

A/B

P

Figure 7.5 134

Example: Suppose V C = 50, and demand is given by Q = 20, 000−100P , that is, A = 20, 000 20, 000 A and B = 100. The price point where demand becomes zero is = = 200. The profit B 100 1 maximizing price is (200 + 50) = 125. 2 1 Note that if V C increases by $1 to 51, P ∗ increases to (200 + 51) = 125.5, that is, P ∗ 2 increases by $0.50. Note 1: The constant elasticity demand curve and the linear demand curve both give good fit to the data if you consider small variations in price from the current level, and you can use either one to determine how a small change in price from the current price level will affect price. Note 2: The two demand functions imply very different demand conditions if you consider a wide range of price variation. The constant elasticity demand function implies that there is a residual demand even when price is very high. That only makes sense if some customers feel they absolutely need the product and would not accept a substitute regardless of price. In contrast, the linear demand function implies that the product has substitutes, and demand becomes zero if price is too high. Note 3: The two demand functions have different implications for profit maximizing price if V C changes. • If V C increases by $1, the profit maximizing price for a constant elasticity demand curve E with E > 1 increases by dollars, which is more than $1. (Meaning: The marketer E−1 should pass along the increased variable cost plus charge a mark-up on it.) • In contrast, if V C increases by $1, the profit maximizing price for the linear demand curve increases by $0.50. (Meaning: The marketer should split the increased variable cost with the customer.) Note 4: The profit maximizing price P ∗ always satisfies equation (7.4). However, for the linear demand function, we cannot use equation (7.4) directly to find P ∗ , because E itself depends on price. Specifically, for the demand function Q = A − BP given by equation (7.7), price elasticity of demand (²) and price sensitivity (E = |²|) are given by: (7.9) ² =

−P , (A/B) − P

that is,

E =

P . (A/B) − P

Thus, E is close to zero if P is small, and very large if P is close to A/B. You can easily verify the following: A • The linear demand function given by (7) is inelastic (E ≤ 1) if P ≤ , and elastic if 2B A A . At P = , E = 1. P > 2B 2B 135

• At P = P ∗ , E =

1 . PV

Note 5 (Double Marginalization): Consider a channel of distribution with multiple levels where each level sets price to maximize its own profit independently. Since each level equates its marginal cost with marginal revenue, there is a mark-up at each level. As a result, the end customer pays a higher price if there are more levels to a channel of distribution compared to the case the channel levels act as a single entity. Double marginalization occurs for both constant elasticity and linear demand functions. However, the specifics are different. Consider a simple two level channel of distribution (manufacturer + retailer) where the manufacturer’s unit variable cost is V C, and the retailer’s unit variable cost is the unit price the manufacturer charges the retailer. Let PM and PR denote the unit price the manufacturer charges the retailer, and the retailer charges the end customer, respectively. Then: • Under the constant elasticity demand function given by equation (7.6): (7.10) PM = V C ∗

³

h E ´ 1 i = V C 1+ , E−1 E−1

Note that the retailer’s unit margin is

PR = V C ∗ (

h E ´2 1 i = PM 1 + E−1 E−1

E times the unit margin of the manufacturer. E−1

• Under the linear demand function given by equation (7.7), 1 A 1 A (7.11) PM = ( + V C) = V C + ( − V C), 2 B 2 B

PR =

3A V C 1 A + = PM + ( − V C) 4B 4 4 B

Note that the retailer’s margin is half the margin of the manufacturer. 7.12.2 Estimation of Price Sensitivity from a Buy-Response Curve: The data shown in Section 11.4 can be used to find the price sensitivity demand. I will discuss both examples. Example 1: I fitted a constant elasticity demand function: K ∗ P E , where E is the price sensitivity, with the data from Example 1 (Rolling Stone Concert). Taking logarithms: ln Q

=

ln K − E ∗ ln P,

where ln is the natural logarithm. I then fitted this equation using regression analysis and got: ln Q = 18.92 − 4.20 ∗ ln P. Hence, the estimated price sensitivity is 4.20. Example 2: I fitted both the constant elasticity demand function and the linear demand function. 136

Results for Constant Elasticity Demand Function: Here, the demand function can be expressed as: Q = KP −E , that is, ln Q = ln K − E ln P. Using regression analysis, I got the following results: Men ln Qmen = 6.95 − 1.04 ln P Women ln Qwomen = 11.5 − 2.61 ln P All ln Qall = 8.48 − 1.38 ln P From these results, the demand for men is barely elastic (E = 1.04) while the demand for women ia fairly elastic (E = 2.61). The demand for the combined market has an intermediate level of price sensitivity (E = 1.38). Results for Linear Demand Function: Here, the demand function is Q = A − BP . Using regression analysis, I got the following three demand equations: Men Qmen = 131 − 1.68P Women Qwomen = 104 − 1.73P All Qall = 119 − 1.70P To help comparison, the demands are in percent form in all three equations. The price sensitivity (E) for men, women, and the combined market (all) are given as follows: • Men: Emen =

P P = (131/1.68) − P 77.98 − P

• Women: Ewomen = • All: Eall =

P P = (104/1.73) − P 60.12 − P

P P = (119/1.70) − P 70 − P

Thus, at any given level of price, women are more price sensitive than men, and the price sensitivity of the combined market is intermediate to the price sensitivities of men and women. For example, at P = $40, the price sensitivities are: • Men: Emen =

40 = 1.05 77.98 − 40

• Women: Ewomen = • All: Eall =

40 = 1.99 60.12 − 40

40 = 1.33 70 − 40

The profit maximizing prices are also different for men and women. For example, suppose V C = 10. Then, the optimal prices for men only, women only, and the combined market (all) are given as follows: 137

Men

∗ Pmen =

i 1 h 131 + 10 = 43.99 2 1.68

∗ Women Pwomen =

All

∗ Pall =

i 1 h 104 + 10 = 35.06 2 1.73

i 1 h 119 + 10 = 40.00 2 1.70

Example 3: I fitted both the constant elasticity demand function and the linear demand function with the data. Results for Constant Elasticity Demand Function: Here, the demand function can be expressed as: Q = KP −E , that is, ln Q = ln K − E ln P. Using regression analysis, I got the following results: Men ln Qmen = 7.71 − 0.757 ln P Women ln Qwomen = 8.18 − 0.899 ln P All ln Qall = 7.91 − 0.821 ln P From these results, all the three demand curves are inelastic. Results for Linear Demand Function: Here, the demand function is Q = A − BP . Using regression analysis, I got the following three demand equations: Men Qmen = 121 − 0.443P Women Qwomen = 110 − 0.433P All Qall = 115 − 0.438P To help comparison, the demands are in percent form in all three equations. The price sensitivity (E) for men, women, and the combined market (all) are given as follows: • Men: Emen =

P P = (121/0.433) − P 273.14 − P

• Women: Ewomen = • All: Eall =

P P = (110/0.433) − P 254 − P

P P = (115/0.438) − P 263 − P

Thus, at any given level of price, women are more price sensitive than men, and the price sensitivity of the combined market is intermediate to the price sensitivities of men and women. For example, at P = $150, the price sensitivities are: • Men: Emen =

150 = 1.22 273.14 − 150 138

• Women: Ewomen = • All: Eall =

150 = 1.44 254 − 150

150 = 1.33 263 − 150

The profit maximizing prices are also different for men and women. For example, suppose V C = 100. Then, the optimal prices for men only, women only, and the combined market (all) are given as follows: Men

∗ Pmen =

i 1 h 121 + 100 = 186.57 2 0.443

∗ Women Pwomen =

All

7.13

∗ Pall =

i 1 h 110 + 100 = 177 2 0.433

i 1 h 115 + 100 = 181.50 2 0.438

Exercise Problems

Scenario for Problems 1-3: Suppose you sell two products, A and B. Product A: Price/unit = $10, variable cost/unit = $5, current demand = 10,000 units, fixed cost = $20,000, (Own) price elasticity of demand = −2.5. Product B: Price/unit = $20, variable cost/unit = $12, current demand = 8,000 units, fixed cost = $30,000, (Own) price elasticity of demand = −2. The cross price elasticity of demand of A with respect to price of B is 0.5. The cross price elasticity of demand of B with respect to price of A is 1.0. 1) Are these two products substitute products or complementary products? 2) Compute each of the following: (a) Profit from A. (b) Profit from B. (c) Profit from A and B combined. 3) Suppose you have increased unit price of A to $10.40, but kept the price of B unchanged. Compute each of the following: (a) Demand for A. (b) Profit from A. (c) Demand for B. (d) Profit from B. (e) Profit from A and B combined. 139

4. Scenario: Suppose you sell two products A and B. The current prices, current demands, costs, and elasticities of demand are as follows: Product A: Price/unit = $10, variable cost/unit = $5, current demand = 10,000 units, fixed cost = $20,000, (Own) price elasticity of demand = −3. Product B: Price/unit = $20, variable cost/unit = $12, current demand = 8,000 units, fixed cost = $30,000, (Own) price elasticity of demand = −2. The cross price elasticity of demand of A with respect to price of B is 1.5. The cross price elasticity of demand of B with respect to price of A is 1.5. 4.(a) Compute (i) the profit from product A, (ii) the profit from product B, and (iii) the combined profit from products A and B at the current prices. 4.(b) Suppose you increased the price of product A to $10.10/unit, but kept the price of product B the same ($20/unit). At these new prices, compute: • The new demand for product A. • The new demand for product B. • The profit from product A. • The profit from product B. • The combined profit from products A and B. 4.(c) Suppose you increased the price of product A to $9.90/unit, but kept the price of product B the same ($20/unit). At these new prices, compute: • The new demand for product A. • The new demand for product B. • The profit from product A. • The profit from product B. • The combined profit from products A and B. 4.(d) Keeping the price of product B unchanged at $20, which of these three prices of product A ($10.00, $10.10, and $9.90) will you choose if you wanted to get the highest profit from product A? 4.(e) Keeping the price of product B unchanged at $20, which of these three prices of product A ($10.00, $10.10, and $9.90) will you choose if you wanted to get the highest combined profit from products A and B? 140

5. Suppose you are organizing a rock concert at the Carrier Dome. Your total fixed cost is $100,00, your variable cost per unit is $20, and the demand Q (number of tickets) is the following linear function of price: Q = 20, 000 − 250P The relevant formulas for the linear demand function are given below. The profit maximizing price here is: ´ 1 ³ 20, 000 1 P∗ = + 20 = (80 + 20) = 50 2 250 2 5.(a) Compute the price sensitivity (E) at each of the following three levels of price: $40, $50, $60. [To find E, the simple way is to increase P by 1%, compute the % change in Q, and compute elasticity ² by dividing % change in Q by % change in P . E is the absolute value of ². You may also use the formula for E given by Equation (7.9).] 5.(b) Compute the PV ratio at each of the following three levels of price: $40, $50, $60. 1 5.(c) Suppose you wish to maximize profit. Comparing E and from 2(a) and 2(b), discuss PV how and whether price should be changed at each of the three price levels $40, $50, and $60. [Should price be changed at all? If yes, then do you increase or reduce price in order to increase profit?]

7.14

Additional Reading

Thomas T. Nagle and Reed K. Holden, Strategy and Tactics of Pricing: A Guide to Profitable Decision Making, third edition, 2002, Prentice Hall. The sections on competitive bidding and factors affecting price sensitivity in the Reader uses concepts presented in this text.

141

8

Selected Topics on Distribution Strategy

8.1

Rationale for Marketing Intermediaries (Middlemen)

Marketing intermediaries (also called resellers, middlemen) exist when they can perform channel tasks more efficiently than the manufacturer. 8.1.1 Examples of Channel Tasks: 1. Research to determine nature and size of demand. 2. Promoting the product to target customers. 3. Contacting the customers. 4. Negotiating the terms of sale (price, delivery, financing, service). 5. Risk bearing: responsibility for unsold items; risk of bad credits. 6. Physical distribution. 7. Smooth discrepancy between needs of producers and consumers: Assortment, spatial, temporal, quantity. Typically, a manufacturer produces a limited number of products in large quantities in order to achieve economy of scale at production facilities at a small number of locations. In contrast, consumers are dispersed spatially and each consumer desires a limited quantity of a variety of products. As a result, we have: • Assortment discrepancy: The customer wants an assortment of different types of products. The manufacturer produces a limited number of products. • Quantity discrepancy: The customer wants a small amount of each type of product. The manufacturer wants to produced a large amount of a homogeneous product to achieve economy of scale. • Spatial discrepancy: The manufacturer produces the products in a small number of locations. Customers are dispersed over a large area. • Temporal discrepancy: This is the discrepancy between when the product is produced and when it is consumed. For example, food products made with fruits or vegetables are often produced seasonally but consumed throughout the year. In contrast, a garment manufacturer may produce winter jackets throughout the year to maintain a stable labor force, but sell its products only in fall and winter. These discrepancies need to be removed. 142

8.2

Reasons for having intermediaries

Sometimes intermediaries can perform the channel tasks more efficiently than the manufacturers. For example: 1. Increased Contact Efficiency: The presence of intermediaries can significantly reduce the number of transactions necessary in an economy of exchange. For example, consider a market with three manufacturers who produce distinct products (bread, cereal, and soft drink) and 3 customers who each buy all three products. If each manufacturer had to interact with each customer directly, 3 × 3 = 9 transactions would be necessary. On the other hand, if an intermediary purchased the products from all three manufacturers and sold them to all three customers, only 3 + 3 = 6 transactions would be necessary. More generally, if there are C customers who each buy products from the same M manufacturers, then M × C transactions are necessary if there is no intermediary. In contrast, (M + C) transactions are necessary if there is an intermediary between the manufacturers and the end customers. Note that contact efficiency provided by an intermediary increases with fragmentation in production and purchase. If one corporate entity produced all products (M = 1) or if only one customer purchased all products (C = 1), there would be no gain in contact efficiency from having a middleman. As M and C increase, the contact efficiency offered by the intermediary, as measured by the difference between M × C and (M + C), increases. While these examples are simplistic because we implicitly assumed that each transaction is equally costly to the economy, the general idea is correct: whenever production and consumption are fragmented, the presence of an intermediary makes the economy efficient by reducing the number of contacts. In addition to reducing the number of transactions necessary, the intermediaries can standardize transaction (e.g., by establishing a market price) and thus make distribution more efficient. M1

M2

M3

- C1 H * © µ ¡ @HH © © ¡ @ HH ©©¡ @ HH © ¡ H©© @ H @ ©© HH ¡ © ¡H ©@ HH ¡ @ ©© H© j H @¡ © * C2 H ©© HH ¡@ © H ¡ @ ©© HH © ¡H ©@ © ¡ H @ © ¡ ©© HH @ H ¡ ©© HH@ ¡© HH @ ©© ¡ j H R C3 @

143

No Retailer Present M × C = 3 × 3 = 9 transactions

M1

HH

HH

© ©©

H

H j

C1

©©

¼ ©

Retailer Present M + C = 3 + 3 = 6 transactions

M2

M3

-

© ©©

©

* ©©

R

C2

¾

YH H

HH

HH

C3

2 Reduction of Discrepancy. An intermediary can often smooth the discrepancy between production and consumption more efficiently than the manufacturer. This increased efficiency usually arises from economy of scale. For example, by carrying goods from several manufacturers in the same truck, an intermediary may be able to reduce the cost of transportation. Note. The importance of the intermediary grows with the discrepancy between the offerings of the manufacturers and the needs of the customers. For example, a large steel manufacturer usually encourages its small volume customers, who typically need a wide variety of steel products, to buy from wholesalers. In contrast, large volume buyers, who typically need homogeneous supplies of steel in large lots, are urged to deal directly with manufacturer. 3 Market Knowledge. By specializing in local market conditions, the intermediary is in a better position to estimate local demand, advertise in local media, offer credit to customers, and provide after sales service.

8.3

Distribution Intensity

A producer has to decide how many sales outlets should carry its products in a given area. Three basic choices are available. 8.2.1 Intensive Distribution. Under this strategy, the product or brand is placed in as many outlets as possible. Convenience goods such as newspapers, candy, cigarettes, deodorants, toothpaste, etc., are usually marketed this way. Marketers of products with a higher price have also used this strategy successfully. For example, personal computer manufacturer Compaq used this strategy to gain access to a large market. Key Features (a) Generally, this strategy is used for convenience products. Typically, the consumers know about the product already and the nature of the sales outlet does not affect how a consumer evaluates the product. Alternatively, purchasing the product represents a low involvement 144

choice from a class of products with low differentiation. Availability of the product is the key to purchase. This is particularly relevant for impulse products like candy bars that are only purchased if the customer sees them. (b) The store itself does not promote the product as promotion is unlikely to benefit the store directly. Also, the producer does not expect the store to provide information about the product or demonstrate how to use it. (c) If an intensive distribution strategy is used, competition among resellers would almost certainly drive down the price of the product at the retail level. The retailers, in turn, would demand the product at a lower price from the manufacturer. Thus, the gross margin/unit would decrease, and the manufacturer would have to sell more units to break even or generate a profit. Advantage: Intensive distribution allows a consumer to make a purchase with the minimum effort. This usually increases sales in the short run. Disadvantages: 1. Costly: (i) More accounts have to be served. Often a manufacturer of a low priced item like a candy bar uses wholesalers to reach small retailers rather than serve them directly. The wholesaler achieves economy of scale by carrying the products of many manufacturers. (ii) The manufacturer now has to bear the cost of promoting the product. 2. An intensive distribution strategy may not be suitable for a brand with a quality image which enjoys a substantial profit margin at present. Sometimes, the image of quality arises out of the exclusive nature of the product, and that image is hurt if the product is widely available. For example, the image of Godiva chocolates will be diminished if it is sold at K Mart. Even when the reputation of the product is well established and does not depend on the nature of the retail outlet, intensive distribution may still not be appropriate. For example, suppose Denon, a maker of high quality stereo receivers, has expanded the number of its outlets to include mass merchandisers, mail order stores, etc. An expansion like this usually comes with a lack of control over the marketing of the product to the ultimate consumers, which results in two major problems : Problem 1: Increased competition among resellers. Some retailers would invariably reduce the price of Denon to attract customers. This will force other outlets to reduce price as well, leading to a general lowering of profit margins. The product will now be less attractive to high end resellers who may drop the product. Problem 2: Increased effort required from the manufacturer. This can occur in two ways: (i) Competition reduces profit margin, and any advertising benefits all competing retailers. The resellers are therefore less willing to promote the product. Consequently, the manufacturer has to spend more money towards advertising the product. 145

(ii) Some of the outlets will not have facilities to service the product after a sale is made. Since warranty business is usually not very lucrative, major resellers who offer such service will be reluctant to handle problems with equipment sold by other concerns. The manufacturer may now have to either give more incentive to resellers to provide after sale service, or develop a service network of its own (GE had to do that). Either option is costly. In addition to the problems listed above, some resellers may find an intensively distributed product to be inconsistent with the store image and hence drop the product. (For example, in 1982, Levi Strauss decided to sell its jeans through J.C. Penney and Sears. Macy responded by dropping Levi Strauss jeans.) Note: The above discussion demonstrates that an intensive distribution strategy is generally suitable for a convenience product, but it may not be suitable for a shopping or a specialty product. 8.2.2 Exclusive Distribution. Here, only one intermediary sells the product in a specific area. For example, a car dealer may receive the exclusive right to sell a particular brand in a given territory. Key features (i) The purchase represents high involvement decision making by the consumer. The consumer is ready to travel to a specific outlet to examine the product. (ii) The consumer usually requires information before she makes a purchase decision. Also, she may require service (e.g., instruction, repair) after the purchase. As a consequence, selling requires a high degree of involvement by the retailer. In exchange for this effort, the retailer is spared competition from other retailers selling the same brand and typically enjoys close cooperation from the producer/distributor. In general, an exclusive distribution strategy is meaningful if a strong selling effort by the intermediary is required. This is often used in the early stages of a product’s life cycle and/or for the marketing of technically complex products. Examples of exclusive distribution: (i) Imported automobiles (e.g., BMW). (ii) Higher priced furniture. (iii) Construction and farm machinery. 8.2.3 Selective Distribution. This strategy is intermediate to intensive and exclusive distribution strategies where a product or brand is placed in a limited number of outlets in a defined geographic area. Typically, this is suitable for shopping goods which may require some selling effort from intermediaries. Consequently, the producer/distributor protects the retailer somewhat from competition, and provides a limited amount of support. Examples: Medium priced electronics equipment like Sony, cosmetics such as Revlon and Estee Lauder. Note: The choice of distribution strategy depends on the nature of the product, the target customers’ knowledge of the product and the buying habits of the target customers. 146

8.4

Vertical and Horizontal Conflict

Each channel member plays a specific role in the channel and specializes in performing one or more functions. Thus, each channel member depends on the others for success. Ideally, the channel members should cooperate to serve the target market efficiently, and thereby maximize profits for the entire channel. In real life, however, channel members rarely take such a global view. Cooperating to achieve overall channel goals would often require members to give up individual goals. Unwillingness to do that often leads to conflict. Note that conflict may not necessarily indicate open hostility between the parties involved. For example, it may simply refer to price competition leading to a reduction in profits of the entities in conflict. Conflict in a channel of distribution usually arises from disagreement on what role each member should play, and how each member should be compensated. Depending on the levels of the channel of distribution the parties in conflict belong to, conflict may be horizontal or vertical. In our discussions, the term channel level refers to a layer of intermediaries that may exist between the producer and the end customer. For example, wholesalers (who sell to businesses including retailers) and retailers (who sell to consumers) represent different channel levels. If there are no marketing intermediaries, we have a direct channel of distribution. 8.3.1 Horizontal Conflict. Horizontal conflict, which is conflict among members at the same level of the channel of distribution, can usually be of two types: (a) Price competition among resellers at the same level of the channel. For example, a Ford dealer may try to steal sales from another Ford dealer through price reductions, or may encroach on another dealer’s sales territory. (b) When one member at a given level of the channel takes a “free ride” on the other members at that level. Example 1: A member of a fast food franchise may sell inferior product and thereby hurt the reputation of all franchisees. Example 2: Consider a product like a Seiko watch which sometimes requires servicing. As servicing is usually not lucrative, some retail outlets do not want to service the product after it is sold, and the burden falls on the remaining retailers, increasing their cost. This problem is compounded by the fact that the retailers which do not service the product also tend to offer the product at a lower price than retailers who do service the product. 8.3.2 Vertical Conflict. Vertical conflict refers to conflicts between different levels of the same channel. For example, a manufacturer may feel that the wholesalers are demanding unreasonably low prices and decide to start selling directly to retailers or ultimate customers. Like horizontal conflict, vertical conflict can also arise in different ways. Some of the common forms of vertical conflict are as follows: (a) Competition for the same market. In a channel of distribution, members at different levels of the channel often compete with each other for the same market. For example, a 147

manufacturer using a hybrid marketing system (that is, multiple channels of distribution) may sell its products through wholesalers to the retailers, but also may sell them directly to the retailers. Thus, there is now conflict between the manufacturer and the wholesalers as they compete to serve the same market. In a variant of the same problem, a change in distribution strategy may lead to vertical conflict. For example, suppose a manufacturer of a high priced line of cosmetics currently uses a selective distribution strategy and sells the product through upscale department stores like Macy’s. The department stores have invested resources to have display booths and trained personnel to sell this particular line of products. If the manufacturer now decides to switch to an intensive distribution strategy, the department stores will no longer be protected from intense competition, and their investment in resources will be wasted. This will definitely lead to conflict between the manufacturer and the department stores that currently carry the product. (b) Disagreement about sharing of resources and duties to be performed: The success of a channel depends on participation of all members of the channel and there is often disagreement regarding who should be responsible for the various tasks. Some examples follow: (i) Disagreement about who should bear the cost of promotion. For example, manufacturers of small appliances often complain that large retailers like K Mart ask them for unreasonably high promotional allowances. (ii) Disagreement about the level of service to be provided to the customers before and after a sale. This problem is compounded by the fact that some retailers offer service and others do not, and this leads to horizontal conflict as well. Organizations like GE provide customer service directly using toll free numbers. Some other organizations require retailers to commit a significant amount of resources to selling the company’s products. For example, before it adopted intensive distribution, Compaq used to sells its personal computers only through retailers who have a staff trained about Compaq products and dedicated to selling them. (iii) Disagreement about the price the manufacturer charges the middlemen. This may result from a difference in the perception of reality as a manufacturer may underestimate the cost a retailer incurs in selling the product. (iv) Disagreement about who should be responsible for unsold merchandize. (c) Incompatibility of goals. Quite often, members at different levels of a channel of distribution have conflicting goals. For examples, franchisees of fast food franchises like McDonald’s and Burger King have to pay a royalty to the franchiser which is proportional to sales volume. Thus, the franchiser is interested in maximizing sales volume. In contrast, the franchisees themselves are interested in maximizing profit. As a larger sales volume is often associated with a larger cost and does not necessarily result in a larger profit, this may lead to a conflict between the goals of the franchiser and the franchisees. For example, when Pillsbury, the franchiser of Burger King, wanted the outlets to add more lines to increase sales (and counter a similar move by McDonald’s), the franchisees initially resisted as more lines represented a 148

greater cost to the franchisees (cost of adding equipment and employing people to serve the additional lines). The case of soft drink bottlers like Coca Cola is also similar to the problem of fast food franchisees. For example, Coca Cola sells its syrup to the franchisees (the bottlers), and is therefore interested in maximizing sales volume. In contrast, the bottlers are interested in profit, which depends on sales volume as well as costs. Note: Some degree of conflict can actually be healthy for the channel. For example, franchisees may monitor the performance of other franchisees and make sure that product quality does not go down. Similarly, the threat that a manufacturer may bypass it may induce a wholesaler not to keep a large margin of profit. In these cases, the conflict benefits the ultimate customers in the form of better product and lower price, respectively. However, too much conflict hurts the performance of a channel of distribution. This makes coordinating the activities of a channel of distribution important.

8.5

Alternative Channel Arrangements

We discuss two most common forms of channels of distribution, the conventional channel, and the alternative to it, called a vertical marketing system (VMS). 8.4.1 Conventional Channels. A conventional channel consists of independent organizations. Each member of the channel is a separate business entity seeking to maximize its own profit, even at the expense of maximizing profits for the system as a whole. This often results in adversarial relationships between members of the channel. The vertical marketing systems described next have evolved in response to these problems. 4.2 Vertical Marketing System (VMS). Here, members of the channel from different levels (e.g., the producer, wholesaler(s) and retailer(s)) act as a unified system. The unification may involve the entire channel or only a part of it. Usually, either one channel member owns the others or franchises them, or has so much power that they all cooperate. Vertical marketing systems can be classified into (1) corporate VMS, (2) contractual VMS, and (3) administered VMS. (1) Corporate VMS: In a corporate VMS, successive stages of production and distribution are owned and operated by one organization. This is also called vertical integration. Vertical integration can be forward, where a channel member closer to production acquires all or part of the channel leading to the ultimate customer. For example, a manufacturer or a retailer may decide to establish its own distribution outlets. Example: Sherwin Williams (manufacturer of paint products). Vertical integration can also be backward, where retailers or wholesalers assume ownership of institutions that normally precede them in the marketing flow of goods and services. Example: Sears obtains 50% of its throughput from manufacturing facilities which it partially owns. 149

(2) Contractual VMS: A contractual VMS consists of independent firms at different levels of production and distribution integrating their programs on a contractual basis to improve the efficiency of the channel. There are three categories of contractual VMS: (a) Wholesaler-sponsored systems, called voluntary chains: Wholesalers may organize voluntary chains of independent retailers to help them compete with large chain organizations. The wholesaler helps the retailers standardize their selling practices. This may include providing expertise in promotion, pricing, store location, store layout, merchandising, financing, and inventory control. The wholesaler buys centrally in large quantities, and the entire channel benefits from the savings that result from that. Example: Independent Grocers Alliance (IGA) (b) Retailer-sponsored systems called retailer cooperatives: Retailers may take the initiative and organize a new, jointly owned business entity to carry on wholesaling and even production. The members buy through the cooperative, and plan their advertising jointly. The profits are shared by members in proportion to their purchases. Non members are sometimes allowed to purchase through the co-op, but they do not share in the profits. Example: True Value Hardware (c) Franchise organizations: Franchising has grown rapidly in recent years. Sales of goods and services by all franchised companies in the USA exceeded $700 billion in 1990 and is projected to be $1.3 trillion in 2010. The reason for the growth of franchises is that a franchiser with a good business idea can expand its business very fast through franchising. As the franchisees typically pay to join the franchise and subsequently pay royalties, the franchiser does not require a huge capital to start with. The potential benefits to the franchisee include a proven business system, a well known name, promotional support, and market research information provided by the franchiser. Franchising has been a major means for the growth of service industries (e.g., Gymboree, the exercise centers for children). A franchise is an agreement between the owner of a business system, the franchiser, and another person or organization, the franchisee, that wishes to use the business system. The franchisee has to agree to meet the operating requirements of the franchiser, and the agreement takes the form of a legal contract. Franchise systems can be classified in two alternative ways: (a) based on who sponsored the system, and (b) whether the franchise agreement deals with the selling of a specific product. The text discussed the first classification. The second one is discussed here. Based on whether a specific product is involved, there are two basic forms of franchising: (i) Product and trade name: Here, the focus is on what is sold. This is a distribution agreement where a franchiser authorizes a franchisee to sell a product line, using the parent company’s trade name for promotional purposes. The franchisee agrees to buy from the franchisersupplier and also abide by specified policies. Examples: Automobile dealers (e.g., Ford, Honda), petroleum (Chevron, Texaco), soft drinks 150

(Coca Cola). (ii) Business format: Here, the focus is on how a business is run. The franchisee receives a proven business format from the franchiser and, in return, agrees to make payments (typically royalties based on sales volume) to the franchiser and conform to policies and standards. In a franchise agreement based on business format, the franchiser does not always supply the franchisee with merchandize. For example, Little Professor Book Centers, Inc., does not sell its books to the franchised stores (which buy the books from wholesalers). The franchiser only provides management assistance. The 7-11 stores offer another example of a franchise system based on business format. Here, the franchisees are free to buy the merchandize from the franchiser (Southland Corporation) or any other vendor. (3) Administered VMS. An administered VMS coordinates successive stages of production and distribution, not through common ownership or contractual ties, but through the size and power of one of the members, often the manufacturer. In an administered VMS, units may exist with different goals, but a mechanism exists for informal collaboration on inclusive goals. Decision making takes place though interaction of channel members. Examples: (a) Kraft has developed a program to administer the allocation of space in the dairy section of supermarkets. Kraft’s power originates from the company’s 60% share of dairy case volume, exclusive of milk, eggs and butter. (b) Wrigley (the maker of chewing gum) controls the checkout merchandising programs in virtually every major retail outlet in the USA. To accomplish this, Wrigley conducts extensive research on the sales of checkout merchandise and serves as an objective source of such data, which includes information on products marketed by Wrigley as well as products (e.g., Nestle chocolates) which compete with Wrigley for checkout merchandise space. Wrigley also provides retailers with superior fixtures to display their checkout merchandise. This is the reason the layout of checkout merchandise looks virtually the same in every store.

8.6

Power in a Channel of Distribution

Channel members often have and sometimes exercise power over one another, both vertically and horizontally. The major sources of power are the following. 1. Reward power: For example, a manufacturer may offer many different types of reward to the retailer to induce the latter to take actions that benefit the manufacturer. Examples of reward power are: 1(a) Protection from competition. For example, the retailer may be allowed to be the exclusive dealer in a given area. In exchange, the manufacturer may require after-sales customer service, share of promotional costs, or exclusive dealing agreement. 1(b) Greater margins 151

1(c) Promotional expenditure in the territory 1(d) Generous return/refund policy 1(e) Employee training 1(f) Prompt delivery 2. Coercive power: This is the opposite side of reward power where one entity threatens to remove rewards such as an exclusive territory. Both reward and coercive powers can be exercised by large entities with adequate resources. 3. Expert power: Here, the power of one channel member over another comes from some form of expertise. Example 3(a) Wrigley Company does extensive research on how to improve the profitability of checkout merchandizing. Most retailers simply follow the Wrigley recommendations when they display their checkout merchandize. 3(b) Traditionally, national manufacturers had more detailed information on product demands in different market areas. However, with the advent of supermarket scanner cards, retailers have more accurate local demand information than manufacturers. This power allows the retailer to demand more resources from the manufacturer. For example, a manufacturer may have to pay a non-refundable “slotting allowance” of about $100,000 to a retial chain secure shelf space for a new product offering. 4. Legitimate power: Here, the power arises from a binding legal contract. For example, in all of types contractual vertical marketing systems, the parties enter contracts regarding how the business will be run. 5. Referent power: Here, the reputation of a party makes it an attractive partner. Again, it can work both ways. For example, many dealers would love to carry Lexus or BMW automobiles, or Bose speakers. Similarly, most fashion designers would love to have the department store Bloomingdale’s to carry its garments.

8.7

Some Legal Issues Related To Distribution Strategy

We will briefly review two distribution policies which sometimes violate the law: (1) exclusive dealing, and (2) tying agreements. Each strategy is limited by the Clayton Antitrust Act, Sherman Antitrust Act, or Federal Trade Commission Act. It is important to realize that none of these three strategies is illegal by itself (legally speaking “per se”). Rather, it becomes unlawful if it (i) substantially lessens competition, (ii) tends to create a monopoly, or (iii) restrains trade. 1. Exclusive dealing agreement: When a manufacturer prohibits its middlemen from carrying products of its competitors, it is engaging in exclusive dealing. This will occur, for example, if GM prohibits one of its dealers from selling Honda products. 152

Through the use of an exclusive dealing agreement, the manufacturer forces the reseller to concentrate on selling its product. Such an arrangement clearly reduces the freedom of choice of the middlemen and the competitors of the manufacturer now have fewer options to reach the ultimate customers. Exclusive dealing is illegal if these problems are severe: (a) When a manufacturer’s sales volume is a substantial portion of the total volume in a given market, the use of exclusive dealing will exclude competitors from a major portion of the market. In that case, exclusive dealing will be considered illegal. (b) When the manufacturer is a large entity, and middlemen are small concerns, the manufacturer’s power is considered inherently coercive. In that case, the use of exclusive dealing will be considered a restraint of trade and therefore illegal. Exclusive dealing is legal in some situations. Court decisions have allowed exclusive dealing: (a) When a manufacturer is trying to enter a market or its total market share is very small. In a case like this, an exclusive dealing arrangement will force the middlemen to concentrate on selling the manufacturer’s product, and this will in fact increase competition in the market. However, when the same manufacturer establishes a firm foothold in the market, the use of exclusive dealing may not be allowed by the law. (b) When equivalent products are available in the market or when the manufacturer’s competitors have access to equivalent dealers. In these cases, the presence of exclusive dealing will not reduce competition and is therefore allowed. In conclusion, it is important to note that exclusive dealing may not be entirely harmful to consumers. For example, by focusing its efforts on selling the product of one manufacturer, a retailer may acquire greater knowledge of that product and thus be able to serve the customer better. 2. Tying agreement: If a manufacturer sells a product to a middleman only if the middleman also buys another (often unwanted) product from the manufacturer, we call that a tying agreement. A tying agreement limits the ability of the reseller to buy the other product (called tied product) from alternative sources. This may reduce the ability of the reseller to compete, and limit the access the competing manufacturers have to the market. Examples: (1) An appliance manufacturer may want the distributors to carry the entire line of products it offers. (2) A studio may let a theater show its hit movie only if also shows a less popular movie released by the studio. (3) A franchiser may insist that its franchisees buy all the supplies from him. Note that unlike exclusive dealing, a tying agreement does not offer any tangible benefit to a customer. As a result, courts rulings have rarely supported the use of tying agreements. While tying agreements are illegal in most cases, there are some situations where they are allowed: 153

(1) When a new company is trying to enter a market. In this case, the use of a tying agreement benefits the fledgling company and therefore enhances competition. However, if the company establishes itself and gains substantial market share, the tying agreement will no longer be legal. (2) When an exclusive dealer or distributor is required to carry the manufacturer’s full line of products but is not prohibited from carrying competing products. (3) When the second (tied) product is essential for the proper use of the first (tying) product. In this case, the tying product and the tied product are effectively inseparable, and a tying agreement is allowed by the law. For example, consider franchise agreements. Here, the tying product is the trademark itself, and the tied product may be supplies and equipment. ( For example, in a famous lawsuit involving the ice cream franchise Baskin Robbins, some franchisees contended that the supplies (the ice cream products sold by the franchiser) were unlawfully tied to the sale of the Baskin Robbins trademark.) If the franchiser can establish that to run the business properly, the franchisees need to obtain equipment and supplies only from him, then a tying agreement is legal. The rationale is, sales of an inferior substitute under the trademark will reduce value of the trademark. There are several cases involving franchises where the courts allowed a tying agreement. For example, in a case involving the Chock Full O’Nuts Corporations, the courts concluded that the franchiser had the right to supply its franchisees with coffee and baked goods in order to ensure the quality of the product. However, in the same decision, the courts allowed the franchisees to buy other products (e.g., French fries, soft drink syrups, napkins, and glasses) from other vendors.

8.8

Channel Coordination

In Section 7.12, we discussed double marginalization in a channel with multiple levels, where each level sets price to maximize its own profit independently. Since each level equates its marginal cost with marginal revenue, there is a mark-up at each level. As a result, the end customer pays a higher price if there are more levels in a channel of distribution compared to an integrated channel where the channel levels act as a single entity. Because of the higher end price, demand is lower, and the overall channel profit is less than it is for an integrated channel. If the channel is integrated, the end customer pays a lower price and the overall channel profit is higher. Thus, marketers often develop mechanisms to make the different levels of the channel act as a single entity. This is called channel coordination. Clearly, a corporate vertical marketing system can achieve channel coordination. However, channel coordination can also be achieved without having a corporate VMS. Two other mechanisms to achieve channel coordination are as follows: (1) Quantity Discount: The manufacturer/wholesaler offers the retailer a quantity discount if demand exceeds a threshold. 154

(2) Two-part Tariff: The manufacturer/wholesaler offers the product to the retailer at marginal cost, but charges a fixed fee. Example: Consider a channel with one manufacturer and one retailer. The manufacturer sells graphing calculators to the retailer at PM /unit, and the retailer sells the calculators to the end customer at PR /unit. The manufacturer has zero fixed cost, and a unit variable cost of $20. The retailer has no costs other than the PM /unit he pays the manufacturer. Demand is given by: Q(PR ) = 100 − PR if PR ≤ 100, and zero if PR > 100 Integrated Channel: Here, we have linear demand with A = 100 and B = 1, and V C = 20. If the channel is integrated, the profit maximizing retail price is given by equation (7.8), that is, 1 100 PR∗ = ( + 20) = 60 2 1 The demand is: Q(PR ) = Q(60) = 100 − 60 = 40 Profit is: Πintegrated = (60 − 20) × 40 − 0 = 1600 Q(PR ) 6

100

H

HH

H HH

H

HH

H

HH

HH

HH

HH

HH

0 0

20

-

100

PR

Uncoordinated Channel with Independent Channel Members: Using equation (7.11), 1 1 A PM = V C+ ( −V C) = 20+ (100−20) = 60, 2 B 2

1 A 1 PR = PM + ( −V C) = 60+ (100−20) = 80 4 B 4

Therefore: Q(PR ) = 100 − 80 = 20 Manufacturer’s Profit (ΠM ) = (60 − 20) × 20 = 800 155

Retailer’s Profit (ΠR ) = (80 − 60) × 20 = 400 Total Channel Profit is: Πuncoordinated = ΠM + ΠR = 800 + 400 = 1200 This is 400 less than the profit for the integrated channel. Channel Coordinated by Quantity Discount: Suppose the manufacturer offers a quantity discount: PM = 60 if Q < 40, and PM = 45 if Q ≥ 40. The retailer has two choices: • Ignore the quantity discount. Then, his unit variable cost is 60, price is 80, and the profit is 400 as in the uncoordinated channel. • Accept the quantity discount and charge PR = 60 to get Q = 40. Then, his profit is (60 − 45) × 40 = 600. Clearly, the retailer will accept the discount, set PR = 60, and get ΠR = 600. The manufacturer’s profit is: ΠM = (45 − 20) × 40 = 1000 Note that the total channel profit is Πcoordinated = ΠM + ΠR = 1000 + 600 = 1600 = Πintegrated Range of discount: Here, the discounted manufacturer price is PM = 45. Actually this could have been any price in a range as long as both the manufacturer and the retailer have at least the profit they get in an uncoordinated channel. Thus, in this example, the manufacturer requires ΠM ≥ 800, that is, (PM − 20) × 40 ≥ 800,

that is, PM ≥ 40

The retailer requires ΠR ≥ 400, that is, (60 − PM ) × 40 ≥ 400,

that is, 60 − PM ≥ 10,

that is, PM ≤ 50

Thus, we could have used any PM between 40 and 50. The exact number is obtained by negotiation and depends on the relative powers of the manufacturer and the retailer. Channel Coordinated by Two-part Tariff: Suppose the manufacturer charges PM = 20 and a fixed fee of $1000. Then, for the retailer, profit is maximized at PR = 60. Demand is 40, and ΠR = (60 − 20) × 40 − 1000 = 600,

ΠM = (20 − 20) × 40 + 1000 = 1000 156

Once again, the total profit of the coordinated channel is same as that of the integrated channel. Range of Fixed Fee: The manufacturer requires ΠM ≥ 800, that is, the fixed fee must be at least 800. The retailer requires ΠR ≥ 400, that is, (60 − 20) × 40 − Fixed Fee ≥ 400,

that is,

Fixed Fee ≤ 1600 − 400 = 1200

Thus, the fixed fee could be any number between 800 and 1200. The exact number is obtained by negotiation and depends on the relative powers of the manufacturer and the retailer.

8.9

Additional Reading

1. Louis W. Stern, and Adel I. El-Ansary, Marketing Channels, fourth edition, Prentice Hall, 1992. This is the standard text on channels of distribution. The concepts presented in this section are largely drawn from this text. 2. Abel Jeuland, and Steven M. Shugan (1983), “Managing Channel Profits,” Marketing Science, 2(3) Shows how channels can be coordinated using quantity discounts.

157

9 9.1

Selected Topics on Promotion Strategy Basic Ideas

There are four forms of promotion. 9.1.1 Advertising. “Any paid form of non-personal presentation and promotion of ideas, goods, or services by an identified sponsor.” Basic features of advertising: 1. Can reach a mass audience simultaneously. 2. Standardized: The same message reaches the entire audience. 3. Control: The sponsor directly controls the message which reaches the audience. 4. The message that can be delivered is limited in scope and cannot contain a large amount of information. As a result, advertising is most suitable for generation of awareness. 5. The audience cannot provide feedback directly. 9.1.2 Personal Selling. “Presentation of a product to a prospective customer by a representative of the selling organization.” Basic features: 1. Typically, it involves an oral presentation. 2. Compared to advertising, personal selling can be directed more easily at qualified customers. 3. The salesperson can customize message to satisfy the information needs of a specific customer. 4. The salesperson directly interacts with the customer and receives immediate feedback. This helps reduce noise in communication. In addition, the salesperson gets valuable insight into the nature of the needs of the customer. This can help a company (i) identify competition, (ii) determine if customers like the products it currently offers, (iii) get directions for improving the products it offers, and (iv) get new product ideas. 5. By interacting directly with the customer, a salesperson would know whether to try to close a sale or stop trying to sell the product. 6. A salesperson can build a long term relationship based on trust and/or mutual dependence with a customer. This makes the selling task easier in the long run. This can happen only if the salesperson keeps the customer’s interests at heart. 7. Personal selling is a company’s most expensive contact tool. In 1983, the average cost of a sales call was $205. In 1981, American companies spent over $150 billion on personal selling (compared to $61 billion on advertising in that year). This makes effective sales-force management a priority for most companies. 8. Compared to advertising costs, the firm has a greater degree of control over personal selling costs. First, in contrast to advertising space which has to be purchased in large blocks, the size of the sales force can be changed by small amounts. Second, the firm has control over the method of compensating the sales-force. For example, if a firm uses straight commissions 158

to pay the sales-force, selling costs become variable rather than fixed and are incurred largely after sales are made. That is desirable for a firm in a weak financial condition. 9.1.3 Sales Promotion. This is a name applied to a variety of techniques which include coupons, contests, premiums, etc. A sales promotion can be directed at ultimate consumers as well as at intermediaries in the channel of distribution. Basic features of sales promotions: 1. Communication: They gain attention and usually provide information that may lead the customer to the product (e.g., “At your participating GM dealers”). 2. Incentive: They often include something (e.g., price reduction) which is of value to the party addressed. 3. Invitation: Usually, they include a distinct invitation to buy quickly. For example, if a buyer for a department store may be offered a lower price if she places an order during a trade show. Examples: (i) 20% off all men’s shirts from October 20 - 22 at Sears. (ii) “. . . Prices this good cannot last for ever, so buy now.” 9.1.4 Public Relations. The text discusses the general concept very well. I will now briefly discuss publicity. Publicity: “Securing editorial space in a medium which will reach the target audience.” Unlike advertising, here the company usually has little control over the message that reaches the public. If it is favorable, publicity has the following advantages over advertising: 1. High credibility: Since the information is published by independent media, the message is more credible to the target audience. 2. Target audience may be caught off guard: Publicity can reach prospects who may consciously avoid advertisements, salespeople, etc. For example, a person who does not look at computer advertisements may read an article about a new chip by IBM published in the New York Times. 3. Usually, publicity costs less than advertising (or personal selling). However, it is not entirely cost free. 4. It can be conducted much faster than an advertising campaign. 5. More information can be released to the media than one can put in an ad campaign. Of course, in order to get editorial space, the news media should consider the news worth publishing. However, publicity has two potential problems: 1. It is not possible to control what is published by the news media. 2. A given publicity item can usually be used only once as news has a very short shelf life. 159

9.2

Selected Topics on Personal Selling

9.2.1 Introduction Personal selling has two primary objectives, (1) convey product and company information to the customer and receive feedback, and (2) persuade the customer to buy the product. In addition, the salesperson, being close to the customer, can help the company determine the customers’ needs and level of satisfaction with currently available products. The roles played by the salesforce and the types of salespeople employed by a company are determined by these objectives. 1. Roles played by the sales force. A sales force acts as an intermediary between the firm and the customer and usually plays the following roles: (i) Conveyer of information (including “missionary sales”). The sales force facilitates communication between the firm and the customer. It conveys information about the company and its product mix to the customer. It conveys information about the product mix sought by the customer to the firm. (ii) Catalyst for the selling process (creative selling). The sales force can create demand by demonstrating how the product offered by the company can satisfy the customer’s needs. This can take two forms: (a) Developing primary demand for a new product category. For example, IBM’s salespeople showed the target customers how a main frame computer would increase productivity. The purchase will be a new task for the customer. (b) Developing selective demand for the company’s product. Here, the sales force has to compare the company’s product with the competitors’, and justify why the customer should choose it. The purchase is likely to be a modified rebuy or a new task for the customer, depending on what (s)he is using at present. In her role as a catalyst for the selling process, a salesperson usually has some control over the product mix she can offer a customer. Thus she can modify the mix to make it more acceptable to a customer. For example, this may involve matching a price cut by a competitor. Other common examples are offering a favorable trade in for an old product, reducing the freight charge, or guaranteeing early delivery. (iii) Service agent. The sales force sometimes performs service related tasks, some of which are listed below: (a) Order processing. (b) Delivering product. (c) Help arrange financing. (d) Service the product after purchase (e.g., arrange repairs, supply parts). (e) Teaching the customer how to use the product. Note: I have often used the term sales force rather than salesperson in the discussions above. This is because the selling may be done by a selling team rather than one salesperson. Also, different salespeople might perform different tasks involved in personal selling. 2. Types of selling jobs. Salespeople may be categorized based on the functions they 160

primarily perform for the company. The main categories are as follows. (i) Order getters. Order getters specialize in new customers and their tasks include: (a) Identifying prospects and their needs. (b) Attracting attention and making the initial sale. This may include setting up displays, giving product demonstrations, performing promotional work, making sales presentations, and handling objections. (c) After the sale, stimulating further demand for the product. (ii) Order Takers. Here, the demand already exists. The salesperson’s tasks primarily include: (a) Taking orders. (b) Providing delivery services. (c) Offering best deals. An order taker can work inside(e.g., a salesperson working for a mail order PC company), or in the field (outside) (e.g., a representative of an apparel manufacturer visiting the buyers of department stores). (iii) Sales Support Personnel. (iii)(a) Missionary. The missionary salesperson is usually not expected or permitted to take an order, and her primary task is to build goodwill or to educate the potential or current user. The absence of selling pressure gives her greater access to potential customers. A missionary salesperson can help the company identify the product mix sought by the customer and determine whether these needs are satisfied by existing products. For example, representatives of text-book publishers routinely visit professors and talk about new text offerings, supply review copies, etc. They also take ideas about the competition and new texts back to the publishers. (iii)(b) Technical salespeople. The technical salespeople have specific technical knowledge and generally come from functional areas such as production and finance. They usually do not make actual sales unless the product is complex and new. A technical expert often has a counterpart in the buying organization. Examples: Example 1. An engineering salesperson who is primarily a consultant to a client company. Example 2. A technical expert who prepares proposals for complex products that will be purchased through bidding. Example 3. A technical expert who assists an order getter handle objections during a sales presentation. Example 4. A financial expert who specializes in making complex deals. In addition, sales support personnel also include people who make deliveries or handle complaints. 9.2.2 Sales Force Organization (Structure). 161

A firm should organize its salesforce to serve its market efficiently. There are four basic structures: (1) geographical, (2) product, (3) customer and (4) function. Larger firms typically use a complex structure, i.e., a combination of two or more basic structures. The Basic Structures: 1. Territorial (Geographic) Structure. This is a very simple sales organization where each salesperson is assigned a specific geographic territory, and she represents all the products from the company in that area. Advantages: (i) A salesperson’s duties are clearly defined. To the extent personal selling makes a difference, the salesperson receives credit or blame for the sales in a territory. (ii) The salesperson has a chance to develop local business and personal ties. This leads to long term relationships with the customers and a better understanding of the local market conditions. This is particularly meaningful when the customers are individual consumers or small companies. (iii) Travel costs are low. This is particularly meaningful when customers are scattered spatially. The geographical structure is the basic sales force structure. However, there are situations where the geographic structure is not appropriate. If the company sells several unrelated products and the selling of each product requires technical expertise, one sales representative may not be able to function efficiently. In that situation, a product oriented sales force may be used. If the customers are heterogeneous, and different customers have different needs, one sales representative may not be able to handle all of them efficiently. In that situation, the company may use a customer-structured sales force. Finally, sometimes personal selling may involve more tasks than generating immediate sales. For example, it may be desirable to educate customers about potential benefits of the product, provide service after a sale is made, train new recruits and try to identify and convert new prospects rather than simply sell to existing customers. When tasks are diverse and need different skills, a firm may structure its salesforce based on function. 2. Product Structure. Sometimes, when a company sells several unrelated, complex products, a given salesperson specializes in selling certain products. Advantages: (i) Each salesperson is more knowledgeable about what (s)he is selling, and can answer customer questions and handle objections better. (ii) If the company uses product management, a product oriented salesforce will be more efficient, since a salesperson has to work with only one division in the company. Potential Disadvantage: To begin with a product organization is more costly in terms of travel expenses than a territorial organization. In addition, a company may have several product divisions, each with its own salespeople, all selling products to the same customers. For example, a hospital supply company may be selling many different products to the same hospitals. In that case, the product-structured sales force may become inefficient due to the 162

following reasons: (i) If the salespeople responsible for the different products do not interact properly, they might compete for the same buyer’s time. (ii) The task of the buyer may become more complex as she now has to deal with several salespeople (rather than one salesperson for a geographically structured salesforce). 3. Customer Structure. When customer needs and purchases vary widely, a company may use a customer oriented sales force where different salespeople or sales teams specialize in different groups of customers. For example, companies often structure their sales forces based on customer type, e.g., major vs minor accounts, and regular vs new customers. The Digital Equipment Corporation (DEC), for example, has a separate sales force for selling personal computers to farmers. Note that a company using a customer oriented sales force is employing market segmentation. Advantage: A salesperson (or selling team) can be more knowledgeable about the needs of a specific category of customers and can therefore sell more effectively to it. Disadvantage: If the customers of a given category are scattered over a wide region, that may lead to high costs of travelling. A special case. Sometimes a customer is so large that the company assigns a salesperson (or selling team) exclusively to that customer. The salesperson is usually responsible for one account with one company for the entire country. This is called national account organization. 4. Functional Structure. This organization is not discussed in the text. Here, the firm organizes its salesforce based on function. For example, some salespeople may specialize in prospecting, some salespeople may specialize in large customers, etc. Clearly, this method makes sense if there are many functions involved and one salesperson or team may find it difficult to master all of these functions. Combination. So far we have discussed four simple salesforce structures. Sometimes a company may use a combination of these structures as well. For example, a salesperson may be responsible for selling a specific product only in a given territory (compare this with the geographical structure where one salesperson sells all products). To illustrate when a complex sales force structure may be useful to the selling company, let us consider business market segmentation based on decision style. (For a more complete description of the decision styles listed here, see Section 3 of the Reader.) Buying organizations can be segmented as follows based on decision style: (1) Bureaucratic Decision Style: Primarily government agencies. Many organizations with trustees (e.g., hospitals, universities) also use this form of decision making. Needs: Meet specifications. Proper paper work. Low prices. Seller’s perspective: Assign clerical staff to learn when the need arises and what paper work is required. May need technical salespeople to meet specifications, prepare bids and write proposals. Usually, there is no need to make sales presentations. 163

(2) Small Companies: These are typically characterized by (i) small order size, (ii) immediate need, and (iii) a small number of decision makers without technical expertise. Needs: Fast delivery. Guarantee that the product will work. After sale service. Seller’s Perspective: A geographic (territorial) salesperson is most appropriate here. The customer will feel comfortable dealing with somebody she can access easily after the product is purchased. (3) Large companies with functional divisions: These are typically characterized by large orders and a long period of planning and comparing alternatives. As similar needs may recur, the buying firm may wish to establish a long term relationship with the seller to avoid lengthy decision making in the future. Seller’s perspective: Need a selling team. Effective (and probably multiple) sales presentations necessary. Need experts to negotiate with and handle objections from experts in the buying team. This can be an ongoing process. If the customer is large, national account organization may be necessary. Team Selling. When the selling task is complex, a company may use a selling team instead of a salesperson acting alone. A selling team may consist of an order getter, technical salespeople to handle objections, support people to arrange deliveries, etc. Team selling makes sense if: (i) The product is complex (e.g., a main frame computer), and the selling team needs experts to answer questions raised by experts in the buying center. (ii) The size of the order is very large. The selling team may require financial experts, experts on the logistics of making deliveries etc. to negotiate the details of the purchase with the buyer. 9.2.3 Salesforce Compensation Selling is a physically demanding job. It is mentally demanding as well since uncertainty is part of life. In addition, the salesperson often has an inferior status compared to the buyer, and she has to confront aggressive, competing salespeople. Clearly, a salesperson needs recognition and reinforcement to maintain a high level of motivation. An enlightened company would provide individual attention to members of its salesforce, providing counselling and remedial training to improve performance, and rewarding superior achievement. Companies use financial as well non-financial incentives to motivate the salesforce. Nonfinancial incentives include peer recognition, paid vacations, etc. The company often ties the earnings of a salesperson to his performance to induce her to work harder. An examination of the common methods of compensating salespeople illustrates that. We first describe three basic methods of compensating a sales force and then discuss two ways to modify the basic forms. Basic Forms: 1. Straight salary. This is appropriate when the job of the salesperson includes primarily non selling duties (e.g., missionary, technical salesperson). Advantage: This provides the salesperson with a stable earning. 164

Problems: (i) The earnings are not related to any performance measure. This method will be effective only if the company can evaluate the quality of the salesperson’s performance and use incentives like social recognition or possibility of promotion (or, threat of losing the job) to motivate her. (ii) A compensation structure based on salary represents a fixed cost for the firm. This may not be appropriate for a firm in a weak financial situation. 2. Commissions. Under this plan, the salesperson is paid a predetermined percentage of the revenue generated. Most companies use a straight commission plan where the salesperson receives a fixed percentage of revenue regardless of the level of sales attained. Some companies employ a sliding commission plan where the percentage depends on the level of sales achieved. A plan based on commissions is appropriate when the amount of sales depends largely on the initiative of the salesperson, e.g., on the number of accounts called. Advantage: There is a clear incentive to work harder. Some of the conditions when use of commissions is appropriate: (i) When the company has a weak financial position. Commission is paid only if a sale is made, and is a variable rather than fixed cost. (ii) When it is hard to monitor how hard a salesperson is working. (iii) When the non selling part of the salesperson’s job is negligible. Disadvantages: (a) The salesperson’s earnings are often unstable. Confronted with a high level of uncertainty in income, salespeople may quit. (b) The salesperson’s objective may be to earn a large amount quickly, and then leave the salesforce. Such a salesperson will try to sell more in the short run at the expense of building a long term relationship with the buyer. This can result in the loss of consumer goodwill for the company. The company should be careful to prevent this. One solution may be to relate the salesperson’s future earnings to the long term success of the firm, e.g., by giving her stocks in the company. 3. Combination of salary and commission: These plans use a salary plus a commission. The commission motivates the salesperson to work harder, and the salary provides a stable income. Modifications: These are used to modify basic forms such as a straight salary or a straight commission plan. 1. Compensation based on sales quotas: Sometimes the incentive a salesperson receives from the firm depends on the attainment of a pre-specified target, also called a quota. The incentives are generally financial. For example, the salesperson may receive a bonus or a higher commission rate if she reaches the target. In some cases, the incentive is non-financial. For example, it may also be purely social recognition, e.g., the color of the uniform a Mary Kay Cosmetics salesperson is allowed to wear. The principal purpose of having a target is to motivate the salesperson to work hard. If the quota is too low, that purpose is not served. If the quota is too high, that is not good either as the salesperson may be discouraged even from trying to achieve it. Clearly, the 165

choice of the target is of central importance to the firm. An enlightened firm should recognize that different salespeople cannot always be expected to generate equal revenues. For example, they may be working in territories with unequal sales potentials, may have different levels of training, etc. When the sales manager is qualified to recognize and evaluate such differences, (s)he may use sales quotas to evaluate the performance of a salesperson. Usually, the sales manager, together with each salesperson, decides on a target sales for a specified period. There is a second reason why a firm may use a quota plan. By aggregating the quotas for the different salespeople, the marketer can obtain an estimate of the total market demand. This estimate can be used to plan production better. In general, the success of a plan using quotas depends critically on the management’s ability to predict sales for a given level of sales effort. 2. Drawing Account: To reduce the uncertainty in a salesperson’s earnings, firms sometimes use drawing accounts where a salesperson can draw a moderate amount of money from the firm to provide herself with a stable earning. For example, the firm may pay the salesperson a fixed amount at regular intervals. Or, a salesperson may be able to draw limited amounts until the outstanding balance reaches a predetermined limit. Typically, a drawing account is used in combination with a compensation scheme based on commissions, and the salesperson is expected to pay back the amount drawn when (s)he makes the sales. In addition to the compensation schemes listed above, firms use other financial incentives to motivate their salespeople. For example, there are sales contests, and the winning salespeople are rewarded. 9.2.4 Sales-force Size: Three methods commonly used to determine sales force size are briefly described. For details, please see “Sales Force Management” by Churchill, Ford and Walker, fifth edition, 1997, Irwin. 1. Breakdown Method: This is a simple approach where the firm estimates its forecasted sales volume s, and the average sales generated by a salesperson p. The size of the sales force is: s n = p For example, if the forecasted sales volume is $5,000,000, and the average productivity is 5, 000, 000 = 20 salespeople. $250,000, you need 250, 000 2. Workload Method: This is a more sophisticated method and has the following steps. • Divide current and prospective accounts into categories. Some firms use the ABC mechanism which says that top 15% of accounts represent 65% of sales volume, the next 20% accounts represent 20% of sales volume, and the last 65% accounts represent 15% sales volume. Some other firms use the 80/20 rule which states that the top 20% of accounts represent 80% of sales. 166

• For each account category, determine the number of calls needed per year, and the length of each call. • For an average salesperson, determine the number of weeks he works in a year, the number of hours he has available in a week, and the fraction of the available time he can spend on his sales calls. From this, determine the average number of hours a salesperson can spend on sales calls in a year. • You can now determine the number of salespeople you need. Example: Acme Company has the following accounts: Account Category A B C

Number of accounts in Category 200 300 1000

Frequency of calls needed once in a month once in 3 months once a year

Number of calls needed in a year 12 4 1

Length of a call 120 minutes 60 minutes 30 minutes

Thus, you need a total of (200 × 12 × 120) + (300 × 4 × 60) + (1000 × 1 × 30) = 390, 000 390, 000 minutes a year, that is, = 6500 hours a year for sales calls. 60 Suppose now an average salesperson works 40 weeks a year, 45 hours a week, and can spend 30% of this time on sales calls (the remainder is spent travelling, preparing reports, arranging delivery, etc.) Thus, an average salesperson has 40 × 45 × 0.30 = 540 hours each year he can spend on sales calls. 6500 Combining, the required size is ≈ 12. 540 3. Incremental Method: In this method, you estimate the revenue and costs as you increase the size of your sales force. Keep on increasing the size as long as the marginal profit from one more salesperson (before considering compensation and other costs related to having another salesperson) exceeds the marginal cost of adding that salesperson. Example: Suppose the marginal cost of adding another salesperson is $100,000, and you have the following information: Number of Salespeople 0 1 2 3 4 5 6

Revenue ($1000) 0 300 580 850 1110 1350 1530

Cost unrelated to to sales force ($1000) 0 120 270 420 570 720 870

Gross Profit

Marginal Profit

($1000) 0 180 310 430 540 630 660

($1000) 0 180 130 120 110 90 30

167

Cost related to sales force ($1000) 0 100 200 300 400 500 600

Marginal Cost ($1000) 0 100 100 100 100 100 100

Note that the marginal profit (column 5) from an additional salesperson exceeds the marginal cost of having another salesperson (column 7) as long as the number of salespeople is 4 or less. Clearly, the net profit after considering sales force related costs is maximized for a sales force of size 4.

9.3

Publicity

Basic Idea: Publicity involves securing editorial space (as opposed to paid space) in media watched, read or heard by the target audience. Objective: The basic objective of publicity is to create a favorable image of the company which will lead to improved performance (sales, investment, etc.) in the long run. Some specific examples of publicity objectives are (i) handling adverse rumors and stories, (ii) building up a good corporate citizen image for the company or industry, and (iii) enhancing the reputation of the company as a pioneer in research and development. In the rest of the section, we consider publicity through the news media only. However, publicity can be conducted by other means also. For example, many nonprofit organizations send newsletters to its target audience (e.g., universities to its alumni) which contain stories about positive developments in the organizations. This method of transmission is a mixture of advertising and publicity, and it is becoming more sophisticated with advances in desk-top publishing. The organization seeking the publicity does not control the message directly. Usually it has to first sell the information to the news media. If that is successful, the media in turn will communicate the information to the ultimate audience. Thus, this is an example of two stage communication. The key to publicity is selling the information to the media. Many organizations have public relations departments to do that effectively. Other organizations hire public relations firms to do the job. One of the tasks of a public relations department is to maintain a good relation with the press. This helps favorable news to get out quickly and also control potential damage from unfavorable events. The public relations department also advises corporate management about public issues and the company’s image. Organizations often use the following two methods to generate publicity : (1) Finding news. Most organizations have something important or interesting happening within it. For example, a member of the research staff may have received a major scientific award, or the organization has started a day care facility for children of the employees. Depending on the magnitude of the news, the organization can use a press conference or press release, or it can submit an article to a magazine. (2) Creating news. A university can host a major conference. A company can start a program to benefit the homeless, etc. Nonprofit organizations generate considerable publicity through events like auctions. The news media usually publishes something only if a significant portion of its audience can personally relate to it. Thus, the type of news an organization would try to get published 168

through a specific medium depends on the nature of the audience the organization intends to reach through that medium. If the ultimate audience is technically sophisticated (e.g., buyers of main frame computers or advanced scientific equipment), news about a technological development may be promoted. Otherwise, the stories often highlight the human element involved. For example, during the early days of space exploration, stories addressed to the popular media focused on the astronauts rather than the technological breakthroughs which occurred at that time. The result was positive for NASA, −− it received the public support and the funds it needed for success. Examples of successful use of publicity: (1) Cabbage Patch dolls: Caleco formally introduced the dolls in 1983 at a Boston press conference where local school children participated in a mass adoption ceremony. The dolls were promoted to child psychologists who publicly endorsed the product. Consequently, the product was recommended by women’s magazines as a perfect Christmas gift, was featured on the Today show, etc. The dolls were sold out by Thanksgiving, and the Cabbage Patch Panic began. (2) 9-Lives Cat Food: Star-Kist Food’s 9-Lives conducted a publicity campaign involving Morris the cat. Among other things, the campaign used: (i) A Morris look-alike contest in nine major markets. Morris made a personal appearance in the shows. These shows generated several news stories. (ii) Publication of “Morris, An Intimate Biography.” (iii) Distribution of a booklet called “The Morris Method” of cat care. The publicity generated by the campaign increased consumer awareness of 9-Lives Cat Food and resulted in a 43% increase in sales.

9.4

Selected Topics on Advertising

9.4.1 Introduction According to the American Marketing Association (AMA), advertising is defined as “Any paid form of non-personal presentation and promotion of ideas, goods, or services by an identified sponsor.” Some numbers: 1) In 1983, U.S. companies spent $75.9 billion on advertising. 2) U.S.A., with 6% of the world’s population, accounts for 57% of world advertising. 3) By the time (s)he finishes school, a typical U.S. high school graduate spends 11,000 hours in the classroom, and 25,000 hours watching T.V.. Before the advent of VCRs with remote controls, an average U.S. high school graduate would have watched 6000 hours of T.V. advertising. 169

4) In the 1970’s, a typical U.S. family would be exposed to over 2000 commercial messages a day. 9.4.2 Media Selection A marketer can choose between many alternative media to reach its target audience. Some criteria for selection are described below: 1. Audience Coverage: To begin with, the marketer has to decide what is the target audience, that is, who are the people she would like to reach. This helps determine how to transmit the message efficiently to them. For example, if you want to advertise a local restaurant, a national advertisement during Olympics will probably reach a very large audience, but it will not be efficient and will cost too much to reach each person in your target audience. The efficiency of audience coverage depends on the following factors: • Selectivity: How selectively can the media reach the target audience? A more selective medium will have less wastage. For example, women’s magazines will be a more selective medium for ads on baby lotions than the New York Times. • Reach: The text defines it as the fraction of a target population that gets exposed to the commercial in a specified time period (usually four weeks), multiplied by 100. For example, suppose you have a target audience of 15 million people, and 6 million out of these 15 million people are exposed to the commercial in a four week time period. Then, 6 × 100 = 40.3 reach is 15 • Frequency: Within the specified time period (usually four weeks), how many times will a receiver be reached by the advertisement if this medium is used? Reach and the frequency are usually multiplied to form a single measure called the gross rating point (GRP) of a commercial. For example, suppose 6 million people within a target audience of 15 million are exposed to the commercial within the specified period and these people are exposed to the commercial 6 × 100 = 40, frequency is 5, and GRP of the five times on the average. Then, reach is 15 commercial is 40 × 5 = 200. 2. Flexibility: A medium is more flexible if an ad can be placed in it at a shorter notice. For example, radio is more flexible than the television. 3. Cost: The marketer should consider the total cost as well as the cost per person reached. In the context of print media, the Cost Per Thousand (CPM) is used to compare alternative media outlets. 3

The idea of reach is simple, but the definition of reach varies from text to text. For example, some texts define reach as the absolute number of people in the target audience that get exposed to the commercial, that is, 6 million in the example above. Some other texts define it as the fraction of the target audience that gets exposed to the commercial, that is, 0.4 in the example above. When you are given these numbers, you should ask how reach is defined.

170

Definition of CPM: CPM =

Cost of Commercial × 1000 Number of target customers reached by commercial

For example, suppose the cost of placing the commercial in a given magazine is $20,000, and the magazine is read by 500,000 men in the age group 35-49, which is your target audience. Then, 20, 000 CPM = × 1000 = 40 500, 000 Note: For a given commercial in a given outlet, CPM varies with your target audience. For example, if your target audience was men in age group 18-34, and 250,000 people in that age 20, 000 × 1000 = 80 group read the magazine, the CPM would be 250, 000 4. The requirements of the message. This depends on the nature of the product and the audience familiarity with the product. For example, the radio may not be the best vehicle for ads about fashion clothing, for which it is desirable that the receiver can see what the product looks like. 5. Editorial Environment. We now evaluate some of the specific media used in advertising using the criteria described above: 1. Television: The television is a relatively new but pervasive medium for advertising. An advertiser would like to find the most efficient way to transmit the message to the target audience. Thus, it is important to know the nature of the audience that watch the different programs, and how many people watch them. This information is available from syndicated sources, e.g., the A.C. Nielsen Company. Nielsen collects data from a panel of respondents using a device attached to the TV which records which channels get turned on and when. From this, Nielsen prepares ratings for TV programs and supplies them to the subscribers. Advantages of TV advertising: • It is a visual medium, suitable for mood advertising, demonstrations, etc. • It can access a large and diverse audience at once. Disadvantages of TV advertising: • Very costly. Hard to transmit information selectively. The advent of cable has reduced the problem. Some marketers are using self selection by receivers to reduce cost. For example, some makers of cosmetics broadcast long (e.g., half hour) commercials using infrequently used cable channels. This works only if the target audience knows where to find the commercials. 171

• Clutter. This problem is major for brands with low market shares as their commercials get confused with those of the market leaders. This problem decreases with an increase in product differentiation. • Viewers often do not pay attention to the commercials, and even switch channels when the commercials come on. In the past, before the days of the remote control, advertisers tried to counter lack of attention from viewers by massive repetitions of the commercials. Now, it is becoming more important to produce commercials that the viewers would like to watch rather than be forced to watch. In general, television commercials should be short and uncomplicated. 2. Radio. Comparison with television: Advantages: • Relatively inexpensive. A commercial can last longer. • More selective: Can reach a local market more selectively. Also, specific radio stations tend to be more specialized in terms of programs they carry. Thus, it is easier to reach a certain type of audience more effectively. • Sometimes the DJ acts as the spokesperson, and members of the audience can call back and generate a dialogue about the product. This generates audience interest. This method has elements of both advertising and publicity. Disadvantage: Visual stimuli cannot be used. 3. Print Media Advantages of print media: • These can reach a specific audience effectively. • Advertisements can provide more information. Visual stimuli can be used. Magazine advertisements sometimes even have small samples or coupons attached to them. • The receiver can retain the message itself rather than a memory of it. This reduces the need for repetition. • In rare cases, marketers have used smell in print media, such as perfume. • By attaching numbered coupons to the print advertisements, the sales response to the advertisement can be measured. Disadvantage of print media: As the medium is static, commercials do not have the impact of television commercials. Types of Print Media: 172

• Newspapers: These are geographically selective and flexible as advertisements can be inserted at a short notice. • Magazines: (i) Consumer Magazines. (ii) Trade Magazines (very selective). Magazine advertising has the desirable feature that people often keep magazines for a long time and pass them around among friends. As a result, reach and frequency both increase. Problem: Magazine advertising can be very costly and is not flexible like newspaper advertising where you can insert an advertisement easily. 4. Outdoor advertising (Billboards): These advertisements are usually stationary and placed where many people will see them. They are often placed next to busy highways. Advantages: • Relatively inexpensive. • They can be placed near a distribution outlet and thus serve the same role as a point of purchase display. Disadvantages: • Clutter. There are often many billboards near one another. • The advertisement cannot contain much information. They are mostly suitable for reminder advertising. • Legal restrictions. In the USA, many communities impose restrictions on how many billboards can be displayed in a given area. • Not selective. 5. Internet: This is growing area and there is intense competition between organizations such as google and Yahoo for advertising dollars. As this is a complex area with evolving pricing strategies, we will not get into this topic in more detail.

9.5

Sales Promotions

9.5.1 Basic Idea Sales promotions include those activities, other than personal selling, advertising, public relations and publicity, that stimulate consumer purchasing and improve the marketing performance of sellers, i.e., the term sales promotion is used to represent a variety of marketing activities. 173

General Comments: 1) Connection to communication: A sales promotion is a crude form of communication which contains elements of advertising and pricing. The message development is usually simple. For example, a coupon usually tells the receiver how much saving is involved, and how to redeem it. Consequently, the decoding is also a simple process. Usually, the only feedback received by the sponsor is the response to the promotion, e.g., number of trials generated. 2) Industry Practice: U.S. companies currently spend a massive amount of money on sales promotion, and estimates range from $ 100 billion to $ 250 billion a year. This is partly a result of reduced effectiveness (too many ads, legal restrictions) of advertising, and the fact that managers now have competence in designing sales promotions. However, the increase in trade promotions have resulted from another factor: the growing power of the resellers. Typically, the manufacturer of consumer goods depend more on major retail chains than vice versa, since the latter carry many types of products and has better knowledge of local market conditions as well. This fact, coupled with increased competition, has forced manufacturers to make concessions to retailers in order to get shelf space. As a result, for many of the packaged food items, trade discounts are becoming the standard practice rather than anything special, i.e., retailers are paying less for the items they carry. 9.5.2 Types of sales promotions and their objectives A sales promotion may be directed at consumers, channel intermediaries, or the company’s own salespeople. Usually, the objective of a sales promotion is to stimulate earlier or stronger market response. In most cases, the effect of a sales promotion lasts for a short time. While a given sales promotion usually lasts for a short time, a company often has a long term strategy of using sales promotions. Examples of Consumer promotions: (1) Price Incentives (i.e., sale): These can be initiated by any member of the channel of distribution, and they involve savings off the regular (i.e., list) price. Price incentives range from cents-off deals for frequently purchased packaged goods to hundreds of dollars off the prices of stereos, televisions, cars, etc. Objective: Usually short term stimulation of sales and inventory clearance. It can also be a form of price discrimination: it induces sales from the price sensitive segment which waits for sale prices, and does not significantly reduce revenue from the price-insensitive segment which would not wait. (2) Coupons: Coupons are certificates which entitle the bearer to a stated saving on the purchase of a specific product. Over 220 billion coupons were distributed in 1988, and about 7 billion were redeemed. Method of distribution: Coupons may be mailed, inserted in a package, or inserted in print advertisements. Short Term Objectives: Coupons have been used to stimulate sales for mature products. Long Term Objectives: 174

(i) Coupons have been used to generate trial of frequently purchased goods in order to shorten the introduction/market development phase. They have been used to induce reuse of frequently purchased products (e.g., 50 cents off your next purchase of Folger’s Coffee), thereby reducing search of alternatives. (ii) The use of coupons is also a form of price discrimination : it induces sales from the price sensitive segment who are ready to do the work required to redeem coupons. (iii) To remind the consumer that the product exists. (3) Bonus Packs: These offer a consumer an additional quantity of the product at no extra cost (e.g., buy one, get one free). These are typically used to stimulate short term sales. Bonus packs have been used by companies as entry barriers. For example, L’eggs distributed bonus packs of its panty hose to prolong the market development stage of No Nonsense, a competing product. (4) Premiums: A premium is merchandize offered free or at a low price to induce consumers to buy a certain product. For example, breakfast cereal manufacturers often offer toys, t-shirts etc as premiums (usually a consumer has to send in a proof of purchase, e.g., a box-top, to collect the premium). Sometimes, the article given as premium is attached to the product purchased (e.g., free matching slippers with purchase of a gown, etc.). (5) Samples: Samples are small quantities of goods which a customer can try free of charge or at a low cost. Methods of Distribution: Samples may be delivered door-to-door or by mail, picked up at stores, or found attached to other products. Objective: Samples are commonly used to induce trial of a frequently purchased packaged good. This can be very effective during introduction. (6) Contests, sweepstakes, and games: Contest → The participant submits an entry to be evaluated by a panel of judges. Sweepstake → The participants submit their names for a drawing. Game → Typically, a consumer gets something (a missing letter, etc) each time she makes a purchase, which can potentially help her win prizes. This form of sales promotion is very effective for every dollar spent by the company since only a fraction of the participants will receive prizes. To participate, it is not always necessary to make a purchase. Even then, increased consumer interest leads to more sales. (7) Point of Purchase Advertising (Displays): These are used to grab a consumer’s attention at the point of purchase. For example, there may be a five-foot cardboard display of the Marlboro man placed above Marlboro cigarettes. The displays offer no price incentive to the consumer. The retailers receive trade discounts for allowing the displays to take place. Note: In methods (1) to (5), every participating consumer receives some benefit. That may represent a major expenditure for a company. Examples of promotions directed at channel intermediaries: 175

(1) Trade Show: Over 5600 trade shows take place every year, drawing about 80 million people. Channel members who place orders during a show receive premiums, discounts, etc. (2) Push money: Manufacturers frequently give financial incentives to channel intermediaries to induce them to push their products. Usually, this involves obtaining increased and/or more prominent shelf space. (For example, a breakfast manufacturer may like to have the product placed at the eye level of children). Push money is also used to induce channel intermediaries to demonstrate the use of products to potential customers. This is most meaningful during the introduction of medium priced appliances (e.g., a new carpet cleaning device). (3) Trade allowances: These are bonuses (cash, extra merchandize, etc.) given to a channel member who buys a certain volume of the product in a given time period. Objective: These will induce the retailers to carry the product during introduction, or in a highly competitive market. These are also used to clear inventory (similar to a sale). Some texts list cooperative advertising as an example of sales promotion directed at the middlemen. Trade promotions are used widely by manufacturers of packaged goods. For example, in recent years, retailers purchased 90% of the packaged food products they carried at a discount from the manufacturers. In comparison, only 15% of the merchandise was sold at a discount to the ultimate consumers. This indicates the power retailers have over manufacturers, and that trade discounts are becoming the industry practice rather than something special. This is probably a reason direct marketing, which bypasses the retailer, is becoming increasingly attractive to manufacturers.

176

10

Estimation of Conjoint Model (Optional Material)

Consider the additive part-worth conjoint model: (10.1) U (X1 , X2 , . . . , Xn )

=

U0 + U1 (X1 ) + U2 (X2 ) + . . . + Un (Xn ),

where U0 is the evaluation score where each attribute is at the lowest level of its range, and each part-worth score is zero if the attribute is at the lowest level of its range. To estimate the conjoint model, we assume that when asked to rate a hypothetical product, the respondent reports the utility given by equation (10.1) plus a random error as the evaluation score Y , that is, (10.2) Y = U (X1 , . . . , Xn ) + ² = U0 + U1 (X1 ) + . . . + Un (Xn ) + ², where ² is the random error. Equation (10.2) can be estimated using dummy variable regression analysis.

10.1

Dummy Variable Regression

We first discuss dummy variable regression. Our focus will be on the interpretation of the coefficients of the dummy variable regression model. Then, we will apply the intuition developed to understand conjoint analysis. I will illustrate dummy variable regression using four examples. Example 1. One Categorical Independent Variable: Consider the population of students in the USA who graduated from college with undergraduate business degrees in 2006 and received jobs. Let Y denote the annual starting salary of a student (unit = $1000). Suppose we can classify the type of the job into one of four mutually exclusive and collectively exhaustive categories: (1) Accounting (2) Finance (3) Marketing (4) Other To represent the four job categories, we need (4 −1), that is, 3 dummy variables. For example, we can use: D1 = 1 if the type of job is Accounting, D1 = 0 if not Accounting D2 = 1 if the type of job is Finance, D2 = 0 if not Finance, D3 = 1 if the type of job is Marketing, D3 = 0 if not Marketing If D1 = D2 = D3 = 0, then the type of job is “Other.” 177

Consider the regression model: (10.3) Y = β0 + β1 D1 + β2 D2 + β3 D3 + ² This equation is simply a way to allow the mean of Y to be different for the four job categories. From (1), Y can be expressed as follows for the four job categories: (1) Accounting: D1 = 1, D2 = D3 = 0 Y = β0 + β1 + ² (2) Finance: D1 = 0, D2 = 1, D3 = 0 Y = β0 + β2 + ² (3) Marketing: D1 = D2 = 0, D3 = 1 Y = β0 + β3 + ² (4) Other: D1 = D2 = D3 = 0 Y = β0 + ² Thus, β0 is the mean of Y over the sub-population of students who received jobs in the “Other” category. β1 = Mean salary of Accounting jobs − Mean salary of “Other” jobs β2 = Mean salary of Finance jobs − Mean salary of “Other” jobs β3 = Mean salary of Marketing jobs − Mean salary of “Other” jobs Example 2. Two Categorical Independent Variables: Let Y , D1 , D2 , and D3 be as defined in Example 1. Suppose, in addition to job type, salary may also depend on the region of the USA where the job is located. Suppose location has four categories: East, West, Mid-West, and South, and is represented by (4 − 1) = 3 dummy variables: D4 = 1 if East, and D4 = 0 if not East D5 = 1 if West, and D5 = 0 if not West D6 = 1 if Mid-West, and D6 = 0 if not Mid-West If D4 = D5 = D6 = 0, then the location is South. Consider the regression model: (10.4) Y = β0 + (β1 D1 + β2 D2 + β3 D3 ) + (β4 D4 + β5 D5 + β6 D6 ) + ² In this case, there are 4 × 4 = 16 combinations of type of job and location. From Equation (9.4), the mean Y for these sixteen combinations are given as follows: 178

Type of Job Accounting Finance Marketing Other

East β0 + β1 + β4 β0 + β2 + β4 β0 + β3 + β4 β0 + β4

Location West Mid-West β0 + β1 + β5 β0 + β1 + β6 β0 + β2 + β5 β0 + β2 + β6 β0 + β3 + β5 β0 + β3 + β6 β0 + β5 β0 + β6

South β0 + β1 β0 + β2 β0 + β3 β0

Note that: • β0 is the average salary of “Other” jobs in the South • For any given location, β1 = Average Accounting salary − Average “Other” salary β2 = Average Finance salary − Average “Other” salary β3 = Average Marketing salary − Average “Other” salary • For any given job type, β4 = Average Salary in East − Average Salary in South β5 = Average Salary in West − Average Salary in South β6 = Average Salary in Mid-West − Average Salary in South In this example, the difference in the mean salaries of two job types is assumed to be same in all locations. Similarly, the difference in mean salaries between two locations is assumed to be same for all job types. In other words, we assume that there is no “interaction” between the effects of the two categorical variables job type and location. Example 3: Suppose now you are rating a wide variety of sandwiches on 0-100 scales (0: strongly dislike, 100: strongly like). The sandwiches are all available at the same price and are similar in quality. They differ on two attributes: (1) Bread: White, Italian, whole wheat, rye (2) Filling: Beef, turkey, ham-and-cheese, egg To represent the four categories of bread, we define (4 − 1) = dummy variables: D1 = 1 if the bread is white, D1 = 0 if not D2 = 1 if the bread is Italian, D2 = 0 if not D3 = 1 if the bread is whole wheat, D3 = 0 if not If D1 = D2 = D3 = 0, then the bread is rye. To represent the four categories of filling, we define (4 − 1) = 3 dummy variables: 179

D4 = 1 if the filling is beef, D4 = 0 if not D5 = 1 if the filling is turkey, D5 = 0 if not D6 = 1 if the filling is ham-and-cheese, D6 = 0 if not If D4 = D5 = D6 = 0, then the filling is egg. Consider the regression model: (10.5) Y = β0 + (β1 D1 + β2 D2 + β3 D3 ) + (β4 D4 + β5 D5 + β6 D6 ) + ² In this case, there are 4 × 4 = 16 combinations of bread and filling. From Equation (39.5), the mean Y for these sixteen combinations are given as follows: Bread White Italian Whole Wheat Rye

Beef β0 + β1 + β4 β0 + β2 + β4 β0 + β3 + β4 β0 + β4

Filling Turkey Ham-and-Cheese β0 + β1 + β5 β0 + β1 + β6 β0 + β2 + β5 β0 + β2 + β6 β0 + β3 + β5 β0 + β3 + β6 β0 + β5 β0 + β6

Egg β0 + β1 β0 + β2 β0 + β3 β0

Note: • When you rate a particular sandwich, your evaluation score may randomly deviate from the mean evaluation score for that sandwich. ² captures such deviations and is zero on the average. • β0 is the average rating of a sandwich of rye bread and egg filling. • For any given filling, β1 = Average Rating of White Bread − Average Rating of Rye β2 = Average Rating of Italian Bread − Average Rating of Rye β1 = Average Rating of Whole Wheat Bread − Average Rating of Rye • For any given bread, β4 = Average Rating of Beef − Average Rating of Egg β5 = Average Rating of Turkey − Average Rating of Egg β6 = Average Rating of Ham-and-Cheese − Average Rating of Egg Once again, we have assumed that there is no interaction between the effects of bread and filling. Alternative Interpretation: 180

• β0 is the mean evaluation score of a “basic” sandwich of rye bread and egg filling. • The term (β1 D1 + β2 D2 + β3 D3 ) is the adjustment to the mean evaluation score for the four levels of the attribute “bread.” – If bread is “rye,” then (β1 D1 + β2 D2 + β3 D3 ) = 0, that is, no adjustment is needed. – If bread is white, then (β1 D1 + β2 D2 + β3 D3 ) = β1 , that is, β1 is the marginal adjustment if bread is white instead of rye. – If bread is Italian, then (β1 D1 + β2 D2 + β3 D3 ) = β2 , that is, β2 is the marginal adjustment if bread is Italian instead of rye. – If bread is whole wheat, then (β1 D1 +β2 D2 +β3 D3 ) = β3 , that is, β3 is the marginal adjustment if bread is whole wheat instead of rye. • The term (β4 D4 + β5 D5 + β6 D6 ) is the adjustment to the mean evaluation score for the four levels of the attribute “filling.” – If filling is “egg,” then (β4 D4 + β5 D5 + β6 D6 ) = 0, that is, no adjustment is needed. – If filling is beef, then (β4 D4 + β5 D5 + β6 D6 ) = β4 , that is, β4 is the marginal adjustment if filling is beef instead of egg. – If filling is turkey, then (β4 D4 + β5 D5 + β6 D6 ) = β5 , that is, β5 is the marginal adjustment if filling is turkey instead of egg. – If filling is ham-and-cheese, then (β4 D4 + β5 D5 + β6 D6 ) = β6 , that is, β6 is the marginal adjustment if filling is ham-and-cheese instead of egg. Example 4: Suppose instead of sandwiches, you are evaluating several hypothetical laptop computers on 0-100 scales (very poor to excellent). These computers are identical in every way except on two attributes: (1) Weight (X1 ): It can be 4 lb, 6 lb, 8 lb, or 10 lb. (2) Hard Drive Size (X2 ): It can be 80 GB, 100 GB, 120 GB, or 160 GB. Even though these two attributes are quantitative, we can treat them as categorical variables with four categories each and represent them with dummy variables. Weight (X1 ): To represent the four levels, we define (4 − 1) = 3 dummy variables D1 , D2 , and D3 as follows: • D1 = 1 if X1 = 6, and D1 = 0 if X1 = 4, 8, or 10. • D2 = 1 if X1 = 8, and D2 = 0 if X1 = 4, 6, or 10. • D3 = 1 if X1 = 10, and D3 = 0 if X1 = 4, 6, or 8. 181

Hard Drive Size (X2 ): To represent the four levels, we define (4 − 1) = 3 dummy variables D4 , D5 , and D6 as follows: • D4 = 1 if X2 = 100, and D4 = 0 if X2 = 80, 120, or 160. • D5 = 1 if X2 = 120, and D5 = 0 if X2 = 80, 100, or 160. • D6 = 1 if X2 = 160, and D6 = 0 if X2 = 60, 80, or 120. Consider the regression model: (10.6) Y = β0 + (β1 D1 + β2 D2 + β3 D3 ) + (β4 D4 + β5 D5 + β6 D6 ) + ² In this case, there are 4 × 4 = 16 combinations of weight and hard drive size. From Equation (4), the mean Y for these sixteen combinations are given as follows: Weight 4 lb 6 lb 8 lb 10 lb

80 GB β0 β0 + β1 β0 + β2 β0 + β3

Hard Drive Size 100 GB 120 GB β0 + β4 β0 + β5 β0 + β1 + β4 β0 + β1 + β5 β0 + β2 + β4 β0 + β2 + β5 β0 + β3 + β4 β0 + β3 + β5

160 GB β0 + β6 β0 + β1 + β6 β0 + β2 + β6 β0 + β3 + β6

This time: • ² is the random error in the reported evaluation score of a hypothetical laptop computer. • β0 is the mean evaluation score of the “basic laptop” computer with weight 4 lb, and hard drive size 80 GB. • (β1 D1 + β2 D2 + β3 D3 ) is the marginal adjustment in mean evaluation score depending on which of the four levels weight (X1 ) is at. – If X1 = 4, then (β1 D1 + β2 D2 + β3 D3 ) = 0, that is, no adjustment is needed. – If X1 = 6, then (β1 D1 + β2 D2 + β3 D3 ) = β1 , that is, β1 is the marginal adjustment if weight is 6 lb instead of 4 lb. – If X1 = 8, then (β1 D1 + β2 D2 + β3 D3 ) = β2 , that is, β2 is the marginal adjustment if weight is 8 lb instead of 4 lb. – If X1 = 10, then (β1 D1 + β2 D2 + β3 D3 ) = β3 , that is, β3 is the marginal adjustment if weight is 10 lb instead of 4 lb. • (β4 D4 + β5 D5 + β6 D6 ) is the marginal adjustment in mean evaluation score depending on which of the four levels hard drive size (X2 ) is at. – If X2 = 80, then (β4 D4 + β5 D5 + β6 D6 ) = 0, that is, no adjustment is needed. 182

– If X2 = 100, then (β4 D4 +β5 D5 +β6 D6 ) = β4 , that is, β4 is the marginal adjustment if hard drive size is 100 GB instead of 80 GB. – If X2 = 120, then (β4 D4 +β5 D5 +β6 D6 ) = β5 , that is, β5 is the marginal adjustment if hard drive size is 120 GB instead of 80 GB. – If X2 = 160, then (β4 D4 +β5 D5 +β6 D6 ) = β6 , that is, β6 is the marginal adjustment if hard drive size is 160 GB instead of 80 GB.

10.2

Back to Conjoint Analysis

Consider a product with 2 attributes: The laptop computer as in Example 4 with attributes weight (X1 ) and hard drive size (X2 ). Suppose we wish to examine how evaluation score changes if X1 varies over the range 4 − 10 lb, and X2 varies over the range 80 GB to 160 GB. Suppose also the evaluation is done on a 0-100 (very poor to excellent) scale. In this case, the additive part-worth model can be expressed as follows: (10.6) U (X1 , X2 ) = U0 + U1 (X1 ) + U2 (X2 ), where U (X1 , X2 ) is the mean evaluation score of a laptop with attribute levels (X1 , X2 ). Without loss of generality, we assume that: • U1 (4) = 0 • U2 (80) = 0 Under these assumptions: U (4, 80)

=

U0 + U1 (4) + U2 (80)

=

U0 + 0 + 0 = U0

Therefore: • U0 is the mean evaluation score of the “basic” laptop computer with each attribute at the lowest levels in the ranges considered, that is, X1 = 4 and X2 = 80. • U1 (X1 ) is the marginal adjustment to mean evaluation score depending on the level of weight (X1 ). If X1 = 4, then no adjustment is needed. • U2 (X2 ) is the marginal adjustment to mean evaluation score depending on the level of hard drive size (X2 ). If X2 = 80, then no adjustment is needed.

183

10.3

Estimation of Additive Part-Worth Model using Dummy Variable Regression

The objective of conjoint analysis is to estimate: (1) The mean evaluation score of the “basic” product, U0 (2) U1 (X1 ) at three or more levels in the range 4 − 10. The extreme levels 4 and 10 are always included. (3) U2 (X2 ) at three or more levels in the range 80 − 160. The extreme levels 80 and 160 are always included. Suppose we selected the following levels of the two attributes: (1) Four levels of X1 : 4, 6, 8, and 10. (2) Four levels of X2 : 80, 100, 120, and 160. Several hypothetical laptop computers are created using different combinations of these attribute levels, and the respondent provides evaluation scores for these hypothetical products. We assume that the evaluation score can be expressed as: (6) Y = U0 + U1 (X1 ) + U2 (X2 ) + ². During estimation, we focus on the four levels of X1 and the four levels of X2 only. As in Example 4, we represent the four levels of X1 by three dummy variables D1 , D2 , and D3 , defined as follows: • D1 = 1 if X1 = 6, D1 = 0 if X1 = 4, 8, or 10. • D2 = 1 if X1 = 8, D2 = 0 if X1 = 4, 6, or 10. • D3 = 1 if X1 = 10, D3 = 0 if X1 = 4, 6, or 8. Again, as in Example 4, we represent the four levels of X2 by three dummy variables D4 , D5 , and D6 , defined as follows: • D4 = 1 if X2 = 100, D4 = 0 if X2 = 80, 120, or 160. • D5 = 1 if X2 = 120, D5 = 0 if X2 = 80, 100, or 160. • D6 = 1 if X2 = 160, D6 = 0 if X2 = 80, 100, or 120. Consider the regression model: (10.8) Y = β0 + (β1 D1 + β2 D2 + β3 D3 ) + (β4 D4 + β5 D5 + β6 D6 ) + ². Proceeding as in Example 4: 184

• β0 = U0 = mean evaluation score of the “basic laptop” with X1 = 4 and X2 = 80. • (β1 D1 + β2 D2 + β3 D3 ) is adjustment to mean evaluation score depending which of the four levels (4, 6, 8, or 10) X1 is at. For these four levels, U1 (X1 ) = (β1 D1 +β2 D2 +β3 D3 ). Thus: – U1 (4) = 0 (by assumption) – U1 (6) = β1 – U1 (8) = β2 – U1 (10) = β3 • (β4 D4 + β5 D5 + β6 D6 ) is adjustment to mean evaluation score depending which of the four levels (80, 100, 120, or 160) X2 is at. For these four levels, U2 (X2 ) = (β4 D4 + β5 D5 + β6 D6 ). Thus: – U2 (80) = 0 (by assumption) – U2 (100) = β4 – U2 (120) = β5 – U2 (160) = β6 When we regress Y against D1 , D2 , D3 , D4 , D5 , and D6 , we get estimates of β0 , β1 , . . ., β6 . Denoting the estimates of β’s by b’s, we get, for a given respondent, the following: • U0 ≈ b0 • U1 (4) = 0, U1 (6) ≈ b1 , U1 (8) ≈ b2 , U1 (10) ≈ b3 • U2 (80) = 0, U2 (100) ≈ b4 , U2 (120) ≈ b5 , U2 (160) ≈ b6

10.4

Estimation of Conjoint Model for Credit Card Data in Section 5.5

The hypothetical credit cards were designed using 4 levels each of three attributes: (1) X1 (interest rate): 6, 9, 12, or 18 percent APR. (2) X2 : (credit limit): 5, 10, 25, or 50 (unit=$1000). (3) X3 (annual fee): 0, 10, 20, or 50. We assumed that the utility for a product (X1 , X2 , X3 ) can be expressed as: U (X1 , X2 , X3 ) = U0 + U1 (X1 ) + U2 (X2 ) + U3 (X3 ). 185

U0 is the utility of the basic product with X1 = 6, X2 = 5 and X3 = 0. U1 (X1 ), U2 (X2 ) and U3 (X3 ) are marginal contributions of the three attributes to utility as their levels increased from the lowest. These are all zero at the basic levels, that is, U1 (6) = U2 (5) = U3 (0) = 0. Each respondent rated 18 hypothetical credit cards, and data from cards 3 and 9, inserted as anchors, were dropped. The remaining 16 cards were re-ranked from 1 to 16. Each card was then given a score of 17−rank, that is, the card ranked 1 got a score of 16, the card ranked 2 got a score of 15, so on. The score of a hypothetical card is the dependent variable Y . Estimation: We use the dummy variable model: Y = β0 + (β1 D1 + β2 D2 + β3 D3 ) + (β4 D4 + β5 D5 + β6 D6 ) + (β7 D7 + β8 D8 + β9 D9 ) + ². The D’s are only meaningful for the attribute levels used in the design, that is, X1 = 6, 9, 12, 18, X2 = 5, 10, 25, or 50, and X3 = 0, 10, 20, or 50. β0 is the utility of the basic product (6, 5, 0). The D’s and the remaining β’s may be interpreted as follows: X1 6 9 12 18

D1 0 1 0 0

D2 0 0 1 0

D3 0 0 0 1

U1 (X1 ) = β1 D1 + β2 D2 + β3 D3 0 β1 β2 β3

X2 5 10 25 50

D4 0 1 0 0

D5 0 0 1 0

D6 0 0 0 1

U2 (X2 ) = β4 D4 + β5 D5 + β6 D6 0 β4 β5 β6

X3 D7 D8 D9 U3 (X3 ) = β7 D7 + β8 D8 + β9 D9 0 0 0 0 0 10 1 0 0 β7 20 0 1 0 β8 50 0 0 1 β9 Evaluating a new product idea: Let B0 , B1 , . . ., B9 represent the parameter estimates for a given person. Then, for a product concept with attribute values within the ranges used in estimation, utility can be estimated as follows: First compute the components: (1) U0 : Use B0 . (2) U1 (X1 ): Compute it as follows: X1 − 6 ). 6 ≤ X1 < 9: U1 (X1 ) = B1 ∗ ( 3

X1 − 9 ). 3 X1 − 12 ). 12 ≤ X1 ≤ 18: U1 (X1 ) = B2 + (B3 − B2 ) ∗ ( 6

9 ≤ X1 < 12: U1 (X1 ) = B1 + (B2 − B1 ) ∗ (

186

(3) U2 (X2 ): Express X2 in units of $1000, e.g., use 15 if credit limit is $15,000. Compute U2 (X2 ) as follows: X2 − 5 5 ≤ X2 < 10: U2 (X2 ) = B4 ∗ ( ). 5 X2 − 10 10 ≤ X2 < 25: U2 (X2 ) = B4 + (B5 − B4 ) ∗ ( ). 15 X2 − 25 25 ≤ X2 ≤ 50: U2 (X2 ) = B5 + (B6 − B5 ) ∗ ( ). 25 (4) U3 (X3 ): Compute this as follows: X3 0 ≤ X3 < 10: U3 (X3 ) = B7 ∗ ( ). 10 X3 − 10 10 ≤ X3 < 20: U3 (X3 ) = B7 + (B8 − B7 ) ∗ ( ). 10 X3 − 20 ). 20 ≤ X3 ≤ 50: U3 (X3 ) = B8 + (B9 − B8 ) ∗ ( 30 Then add: U (X1 , X2 , X3 ) = U0 + U1 (X1 ) + U2 (X2 ) + U3 (X3 ). When you compare multiple new product concepts, a given person will select the option that yields the highest utility.

10.5

Estimation of Conjoint Model for PC Data from Section 5.6

The hypothetical personal computers were designed using 4 levels each of three attributes: (1) X1 (warranty length): 1, 2, 3, or 4 years. (2) X2 : (RAM): 1, 2, 3, or 4 (unit= 128 MB). (3) X3 (hard drive size): 1, 2, 3, or 5 (unit = 20 GB). We assumed that the utility for a product (X1 , X2 , X3 ) can be expressed as: U (X1 , X2 , X3 ) = U0 + U1 (X1 ) + U2 (X2 ) + U3 (X3 ). U0 is the utility of the basic product with X1 = 1, X2 = 1 and X3 = 1. U1 (X1 ), U2 (X2 ) and U3 (X3 ) are marginal contributions of the three attributes to utility as their levels increased from the lowest. These are all zero at the basic levels, that is, U1 (1) = U2 (1) = U3 (1) = 0. Each respondent rated 18 hypothetical personal computers, and data from computers 1 and 6, inserted as anchors, were dropped. The remaining 16 products were re-ranked from 1 to 16. Each PC was then given a score of (17−rank), that is, the PC ranked 1 got a score of 16, the PC ranked 2 got a score of 15, so on. Two models were used to estimate the model: the basic (dummy variable) model, and the ideal point model. These models are now described separately. Estimation: We use the dummy variable regression model: Y = β0 + (β1 D1 + β2 + β3 D3 ) + (β4 D4 + β5 D5 + β6 D6 ) + (β7 D7 + β8 D8 + β9 D9 ) + ². 187

The D’s are only meaningful for the attribute levels used in the design, that is, X1 = 1, 2, 3, 4, X2 = 1, 2, 3, or 4, and X3 = 1, 2, 3, or 5. β0 is the utility of the basic product (1, 1, 1). The D’s and the remaining β’s may be interpreted as follows: X1 1 2 3 4

D1 0 1 0 0

D2 0 0 1 0

D3 0 0 0 1

U1 (X1 ) = β1 D1 + β2 D2 + β3 D3 0 β1 β2 β3

X2 1 2 3 4

D4 0 1 0 0

D5 0 0 1 0

D6 0 0 0 1

U2 (X2 ) = β4 D4 + β5 D6 + β6 D6 0 β4 β5 β6

X3 1 2 3 5

D7 0 1 0 0

D8 0 0 1 0

D9 0 0 0 1

U3 (X3 ) = β7 D7 + β8 D8 + β9 D9 0 β1 β2 β3

Evaluating a new product idea: Let B0 , B1 , . . ., B9 represent the parameter estimates for a given person. Then, for a product concept with attribute values within the ranges used in estimation, utility can be estimated as follows: First compute the components: (1) U0 : Use B0 . (2) U1 (X1 ): Express X1 in years. Compute U1 (X1 ) as follows: 1 ≤ X1 < 2: U1 (X1 ) = B1 ∗ (X1 − 1). 2 ≤ X1 < 3: U1 (X1 ) = B1 + (B2 − B1 ) ∗ (X1 − 2). 3 ≤ X1 ≤ 4: U1 (X1 ) = B2 + (B3 − B2 ) ∗ (X1 − 3). (3) U2 (X2 ): Express X2 in units of 128 MB, e.g., use 1.5 if RAM is 192 MB. Compute U2 (X2 ) as follows: 1 ≤ X2 < 2: U2 (X2 ) = B4 ∗ (X2 − 1). 2 ≤ X2 < 3: U2 (X2 ) = B4 + (B5 − B4 ) ∗ (X2 − 2). 3 ≤ X2 ≤ 4: U2 (X2 ) = B5 + (B6 − B5 ) ∗ (X2 − 3). (4) U3 (X3 ): Express X3 in units of 20 GB. For example, use 2.5 if hard drive size is 50 GB. Compute U3 (X3 ) as follows: 1 ≤ X3 < 2: U3 (X3 ) = B7 ∗ (X3 − 1). 2 ≤ X3 < 3: U3 (X3 ) = B7 + (B8 − B7 ) ∗ (X3 − 2). 188

X3 − 3 ). 2 Then add: U (X1 , X2 , X3 ) = U0 + U1 (X1 ) + U2 (X2 ) + U3 (X3 ). 3 ≤ X3 ≤ 5: U3 (X3 ) = B8 + (B9 − B8 ) ∗ (

When you compare multiple new product concepts, a given person will select the option that yields the highest utility.

10.6

Estimation of Ideal Point Model

As discussed in Section 5.7.3, the ideal point model is an additive part-worth utility model and can therefore be estimated using dummy variable regression. However, it can also be estimated more directly as discussed here. From Section 5.7.3, the utility of (X1 , X2 , . . . , Xn ) can be expressed as: h

U (X1 , X2 , . . . , Xn ) = K −

n X

i

Wi Xi∗ 2 +

i=1

n X

(2Wi Xi∗ Xi − Wi Xi2 ),

i=1

where each Xi is expressed as the difference of the attribute level from the lowest level. Also, as shown in Section 5.7.3: U0 = K −

Pn

i=1

Wi x∗i 2

U1 (X1 ) = (2W1 X1∗ )X1 − W1 X12 .. . Un (Xn ) = (2Wn Xn∗ ) − Wn Xn2 Let Y be the rating score of the hypothetical product (X1 , X2 , . . . , Xn ). Assuming that the respondent reports the utility plus a random error, we have: Y = β0 + (β11 X1 + β12 X12 ) + . . . + (βn1 Xn + βn2 Xn2 ) + ², where: β11 = .. .

Pn

∗2 i=1 Wi xi (Utility (2W1 X1∗ ), β12 = −W1

β0 = K −

of product with each attribute at lowest level)

βn1 = (2Wn Xn∗ ), βn2 = −Wn This regression model can be estimated by regressing Y with X1 , X12 , . . ., Xn , Xn2 . Note: The regression model is actually more general than the ideal point model. Consider any attribute Xi . • If βi1 > 0 and βi2 < 0, we get the classic ideal point model where Wi = |βi2 |, and 2βi1 . Xi∗ = |βi2 | • If βi1 < 0 and βi2 < 0, then the attribute level should be set at the lowest point of the range. 189

• If βi1 > 0 and βi2 > 0, then the attribute level should be set at the highest point of the range. Estimation of Credit Card data: We estimate the regression model: Y = C0 + C1 (X1 − 6) + C2 (X1 − 6)2 + C3 (X2 − 5) + C4 (X2 − 5)2 + C5 X3 + C6 X32 + ², where X1 , X2 , and X3 are same as in Section 5.5, and C0 is the rating of the card with X1 = 6, X2 = 5, and X3 = 0 on a 0-16 scale. The results are provided in the Excel file “creditf01 ideal point.xls” posted as a Blackboard course document. Estimation of PC data: We estimate the regression model: Y = C0 + C1 (X1 − 1) + C2 (X1 − 1)2 + C3 (X2 − 1) + C4 (X2 − 1)2 + C5 (X3 − 1) + C6 (X3 − 1)2 + ², where X1 , X2 , and X3 are same as in Section 5.6, and C0 is the rating of the PC with X1 = 1, X2 = 1, and X3 = 1 on a 0-16 scale. The results are provided in the Excel file “pcf01 ideal point.xls” posted as a Blackboard course document.

190

11 11.1

Questionnaire Construction (Optional Material) Basic Steps in Questionnaire Construction

Data collected from customers can be divided into the following blocks of information: Non behavioral information: (a) State of being (demographics, ownership status). (b) State of the mind (awareness, attitude, satisfaction, etc.). Behavioral information: (a) Product usage, past and present. This is appropriate for frequently purchased products as such information is then available. (b) Media habits. (c) Purchase intentions. This is appropriate for new product ideas, future purchase intentions, and durable products for which data on past purchases are difficult or even impossible to obtain. To begin questionnaire construction, first determine precisely what information you need to gather. Usually, the focus of a questionnaire is current or future usage behavior. The rest of the information should be relevant to this usage behavior. For example, suppose you wish to study a college student’s usage of cellular telephone service. Then, the appropriate state of being variables are the ones that can possibly affect cell phone usage behaviors. The media habits should focus on the media where the cellular service providers broadcast their commercials. Proceed as follows: 1. For each block of information, identify precisely what information you need to gather. It is useful to first create an outline and build the questionnaire around the outline. 2. For each question, select the form of the question (open ended, multiple choice, or combination). 3. Determine question sequence: (a) Start with easy, general, interesting questions. (b) Make the questions gradually more specific. The earlier questions should not sensitize the respondent to what comes later. (c) Keep sensitive questions for the end. Note that what is sensitive varies from population to population. For undergraduate students, age and income are usually not sensitive issues. For working professionals, they often are.

191

Example: Typical Question Sequence for Student Subjects: 1. Demographics: gender, class, place of residence, etc. 2. Ownership status: car, meal plan, etc. 3. Usage of product category: Does the respondent use product at all? If yes, then how often, and how much each time? 4. Value system: How important are different product features to the respondent? 5. Specific product: awareness, usage, rating of the product on relevant attributes. Note: If you may like to place awareness questions on a new page to avoid biasing the respondent. 6. Media habits.

11.2

Different Question Forms

There are three alternative question forms, (1) open ended questions, (2) multiple choice questions, and (3) combinations of open ended and multiple choice questions. We discuss them separately. 11.2.1 Open Ended Questions: Open ended questions do not ask a respondent to select from a list of possible responses. However, they do vary in terms of how much structure they impose on the answers. Consider the following examples with progressively greater structure: Example 1: Why did you join the MBA Program at Syracuse University? Example 2: List below the reasons you joined the MBA Program at Syracuse University. Example 3: What are the most important reasons you joined the MBA Program at Syracuse University? (List up to three reasons.) Reason 1: Reason 2: Reason 3: Example 4: What is the one most important reason you joined the MBA Program at Syracuse University? Reason: Example 5: Name the one MBA program that you most wanted to join when you applied to MBA programs. Name of Program: 192

Note that with increased structure, it becomes easier to code and enter the data. However, with increased structure, the question becomes less flexible. Advantages of open ended questions: The respondents are not constrained by a predetermined set of answers. This flexibility makes open ended of questions appropriate for exploratory studies. Disadvantages: • Depending on how much freedom we allow the respondent, recording the information may be difficult. For example, if a telephone interview is used, the interviewer may introduce errors while recording the answers. • Coding the response may be difficult. • More articulate respondents get recorded more often while less articulate respondents tend to keep silent. This may introduce a non-response bias by over-representing the views of the more articulate respondents. 11.2.2 Multiple Choice Questions: A multiple choice question provides the respondent a list of categories, and asks her to choose one category from the list. Multiple choice questions are commonly used in survey research because of the following advantages: 1. The responses are easy to code and record. 2. It is easier to obtain cooperation from respondents because the task of responding is simpler. Requirements of a multiple-choice question: The categories provided in a multiple choice question must meet three stringent criteria: 1. The categories must be mutually exclusive, that is, exactly one category must apply to a given respondent. For example, the categories in the following question are not mutually exclusive. How often do you go to grocery stores? (Check one) 0-2 times/week 3 or more times/week

5 or more times/week”

2. The categories must be collectively exhaustive, that is, cover all possible responses. You may need to conduct an exploratory study (e.g., focus group) to determine an adequate set of responses. Even then, we may need to include options like “other” in the question statement. 3. The categories must be appropriate, that is, all the respondents should not belong to the same category. For example, suppose we asked: “How much money do you spend on food each month? 193

Less than $5

$5 to $10

More than $10”

Here, virtually everybody will check the category “ not obtain any useful information.

More than $10,” and the question will

To address this problem, we should first identify very high (but still possible) and a very low (but still possible) value of the variable. These numbers define the two end categories. Then, we divide the range into equal parts to obtain the other categories. For example, suppose we determined that $100 is a very low monthly expenditure on food for a student, and $400 is a very high monthly expenditure. We can then prepare an eight point scale as follows: Approximately how much money do you spend on food each month? (Check one) $100 or less $201−$250 $351−$400

$101−$150 $251−$300 Over $400

$151− $200 $301−$350

Example 6.2: Examples of Multiple-Choice Questions: 1. During an average month, how often do you go out for dinner? (Check one) 0-1 times/month

2-3 times/month

7-10 times/month

11 or more times/month

4-6 times/month

2. Do you subscribe to an internet service provider? Yes

No

Note: This is a special case of the multiple-choice question with only two categories. It is called a dichotomous question. 3. Which of the following movies have you seen? (Check all that apply) Borat Cars

Shrek 2 The Da Vinci Code

Pirates of the Caribbean

Note: This is not a single question, but a series of dichotomous questions. In effect, we are asking separate dichotomous questions such as: “Have you seen Borat? Yes No” 11.2.3 Combination of Multiple-Choice and Open-ended Questions: When we do not have the prior knowledge to specify all the categories, we add an “other” category to a multiple choice question. If we wish to know what “other” means, we ask the respondent to specify “other.” This incorporates an element of an open ended question to a multiple choice question. The following two are examples of such combination questions. Example 1: Which state are you from? (Check one) 194

California

New York

Pennsylvania

Florida

Illinois

Other(Specify:)

Example 2: Which newspapers have you read during the last month (check all that apply)? New York Times Wall Street Journal Other(s) (Please list):

11.3

Daily Orange USA Today

Common Mistakes in Questionnaire Construction

Basic Errors: 1. Does the question apply to the respondent? We need to address this issue in two ways. 1.(a) If the issue is confined to just one question, add a category that allows the respondent to tell you that the question does not apply to her. Example: What brand of beer do you like to drink most? (Check one) Budweiser

Coots

Corona

Rabat

Miller

Sam Adams

I don’t drink beer 1.(b) If you are rating an object on several attributes, add a “don’t know” category to each scale. Example: Please rate the Wegmans grocery store on each of the following: Store Hours Selection of items Price

Very Poor Excellent 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7

Don’t Know DK DK DK

1.(c) If an entire block of question do not apply if the answer to a question is “no,” use a screening question that allows the respondent to bypass this block of questions. Example: “Do you smoke cigarettes?” Yes (If yes, please continue with question 10.) No (If no, please skip to question 20.) 2. The words used in the question may be unfamiliar to the respondent. Whenever possible, we should use simple language and terms the respondent can understand. Even apparently simple terms may be unclear depending on the context. For example, it is doubtful if one can answer: “How many milligrams of salt do you like to have in a serving of potato chips?” 195

3. In a multiple choice question, the categories may not be mutually exclusive and collectively exhaustive. In addition, the categories may not be appropriately defined in that all the responses fall in just one category. 4. Lack of specificity. This occurs when terms in a question may be interpreted differently by different respondents. Example 1: This may occur if we use terms like “often” and “regularly” without defining what we mean. For example, instead of asking “Do you watch the CBS Evening News regularly?”, we may like to ask: “During an average week, how often do you watch the CBS Evening News? 0 times

1−2 times

3−4 times

5 or more times

Example 2: This happens when you do not label the end points of a scale. For example, suppose you asked: “Please rate the Wegmans grocery store on each of the following: Store Hours 1 Selection of items 1 Price 1 Some respondents will

2 3 4 2 3 4 2 3 4 interpret 1

5 5 5 as

6 7 DK 6 7 DK 6 7 DK excellent, and others will interpret 7 as excellent.

Example 3: This happens when an object has multiple dimensions, and you ask someone to rate the object without specifying the dimension. For example, suppose we asked: “Do you like grapefruit juice?” It is possible that the respondent does not like the taste of grapefruit juice, but likes the nutritive value of grapefruit juice. 5. Do not ask a question in a way that will require a respondent to do a complex estimation. For example, if we ask somebody how much money he spends at bars in a month, he may (a) think of how many times he goes to bars in a month, (b) how much he usually spends each time, and (c) multiply the two numbers. In this case, it is better to ask two separate questions: (i) How many times do you go to bars in a month? (ii) How much do you usually spend each time you go to a bar? 6. Avoid double-barreled questions. Examples of double barreled question are: (a) Do you like the taste and nutritive value of grapefruit juice? Yes

No

(b) Name the one store that has the best price and selection of DVD movies. To correct a double barreled question, we should ask two separate questions. More Complex Errors: 1. Biasing questions. Here, the question is such that the respondents think there is a preferred answer and gives that answer even when it does not reflect his true feelings or behavior. This 196

problem is more likely to be present when a human contact is involved in data collection (e.g., telephone interview, household interview, etc.). There are two specific cases where this problem is likely to exist. The first occurs when you are trying to find out how many people are interested in a new product idea. Here, the respondents often feel compelled say that they are interested in the new product because they think that is the desired answer. The second relates to behavior which has either socially desirable (e.g., donation to charities, exercise habits) or undesirable (e.g., smoking) connotations. In order to reduce this problem, it is useful to show the respondent that it is okay to give a socially undesirable answer. For example, before asking: “Have you donated blood during the last one year,” we may add the statement: “Due to health and other concerns, many people do not donate blood.” To make the statement even stronger, we may replace many people by most people. The added statement is called a “counter-biasing” statement. Even when the respondent is not deliberately giving you a biased response, the question wording or sequence may generate a biased response. We now describe four cases where that happens. The four factors discussed here are very similar and are likely to be significant only if the respondent is not familiar with issues involved. 2. Order/position bias. For example, if we give a respondent a list of five alternative hypothetical products and ask him which one he likes most, he will tend to choose the ones at the beginning (primacy effect) or at the end (recency effect) of the list rather than the ones at the middle. This problem is greater if the alternatives are very similar from the perspective of the respondent. Also, this problem is more likely to be present in a telephone survey compared to a mail survey because in a telephone survey, the respondent does not have the alternatives in front of him when he makes the choice and must rely on his memory. 3. Categories provided in a multiple choice question may create the response. Consider the following two examples. Example 1: What do you consider to be appropriate price for a 10 megapixel SLR digital camera? (Check one) $400 or less

$401-$600

$801-$1000

$1001 or more

$601-$800

Here, a respondent unfamiliar with digital cameras will tend to check a middle category, leading to a response created by the categories presented. This problem can be addressed by using an open ended question along with a “don’t know” option. Example 2: If you can get a gift coupon to dine at any restaurant in Syracuse, which restaurant will you choose? (Check one) Acropolis Other (specify:)

Olive Garden

Panda West

197

Pizzeria Uno

Unless the respondent has strong favorite, (s)he will be more likely to choose a restaurant listed in the question than a restaurant not included in the list. Thus, it is appropriate not to provide a list at all. 4. Leading question. Consider the following alternative ways of asking the same question: (a) Do you support tax cuts which will lead to the closing of high schools in the New York State? versus (b) Do you support a tax cut that will put money in the hand of the consumer and promote growth? If the respondent does not have a clear opinion beforehand, the two question formulations will generate very different answers. Note that here the object concerned (that is, proposed tax cut) has a favorable and an unfavorable aspects. By focusing on only one aspect of the object, the question creates a biased response. 5. Unclear concept of the status quo. Consider the following wordings of the same question: (a) If medical procedure X is used, 30% of the patients die. Do you support the use of procedure X? versus (b) If medical procedure X is used, 70% of the patients survive. Do you support the use of procedure X? In version (a), the status quo is “the patient will live otherwise.” In version (b), the status quo is “the patient will die otherwise.” Unless the respondent is familiar with the topic, the question presentation will affect her response. 6. Implicit alternatives. Consider the following alternative ways of asking the same question: (a) Do you support a measure to repair highways and schools in New York state? versus (b) Do you support a measure which will raise state taxes by 5 % to repair highways and schools in New York state? Here, the alternatives are (i) tax and improvement, and (ii) no tax and no improvement. In version (a), the alternatives are not clearly specified (i.e., they remain “implicit”), and the respondent may think he can get something without making a sacrifice. In contrast, the alternatives are clearly specified in version (b).

11.4

Itemized Rating Scales

Itemized rating scales are used extensively in survey research. An itemized rating scale is a multiple choice question with a fixed number of categories where the categories represent 198

progressively higher or lower values of an attribute. I now give examples of a special case of itemized rating scales called semantic differential scales where the two end points of the scale are verbally described using words of opposite meaning. Example 1: To rate an object on one or more attributes. Please rate the Whitman School iMBA Program on each of the following by circling your response. Tuition Charged Course Selection Academic Reputation

Very Poor Excellent 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7

Example 2: To rate importance weights of attributes. How important were the following to you when you selected an MBA Program? (Please circle) Tuition Charged Course Selection Academic Reputation

Not important at all Very Important 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7

Example 3. Degree of agreement with a statement. (This is called a likert scale.) Please indicate how strongly do you agree or disagree with the statement, “The Whitman School iMBA Program gives me good value for my money,” by circling your response. Strongly Disagree 1 2 3

4

Strongly Agree 5 6 7

Example 4. Interest in a new product concept. How interested are you in subscribing to a door-to-door grocery delivery service that charges $5 for each delivery? Not interested at all 1 2 3 4 5

Very interested 6 7

Example 5. Customer satisfaction. How satisfied with your current cellular telephone service provider? Very Dissatisfied 1 2

Very Satisfied 3 4 5 6 7

Example 6. Comparison of an object to a reference object on an attribute. Compared to US Air, how much would you rate American Airlines on baggage handling? Much worse 1 2

3

4

Much better 5 6 7

In all these examples, I used seven-point scales which are the most common scales used. However, 4-point, 5-point, and 10-point scales are also used quite often. You should always 199

define the two ends of the scale.

11.5

Coding a Questionnaire

After you prepare a survey, you have to develop a “coding scheme” to convert responses into numbers suitable for data analysis. You need to do the following: 1. Make a complete list of all the different variables you are measuring with the questionnaire. 2. For each variable we identified in step 1, make a complete list of different values it can take. Assign a number to indicate each value the variable may take. 3. Give an ID number to each completed questionnaire. 4. Give a name to each variable. Standard statistical packages allow alphanumeric names of up to eight characters as a variable name. This means the name has to start with a letter, followed by any combination of letters or numbers. For example, X1 , V AR1 , GEN DER, are all valid variable names. People select variable names based on personal taste. I choose variables names based on which item in the questionnaire the variable comes from. For example, X1 is the variable from Question 1, X2 is the variable from Question 2, etc. If Question 3 has three parts (a), (b), and (c), then I use variable names X3a , X3b and X3b . 5. Preparing EXCEL Worksheet: Save the data as an EXCEL worksheet, which can then be used with all major statistical packages. Prepare the worksheet as follows. • The first line of the data should be ID, Variable Name 1, Variable Name 2, etc., in the successive cells. For example, it may be: ID, X1 , X2 , X3a , X3b , X3c . • Each of the following lines comes from one completed questionnaire. Please enter the ID number followed by the numerical values of the successive variables in the successive cells. In the completed worksheet, the first column will be ID followed by the ID numbers of the different questionnaires. Each of the following columns will have a variable name followed by the numerical values of the variable for the different completed questionnaires. 6. Non-Response Code: If non-response occurs for a variable, you have two options: • You may enter a period sign (.) in the cell. The period sign (.) is as a default nonresponse code for the statistical packages Minitab and SAS. • You may leave the cell blank. A blank cell serves as a default non-response code for the statistical packages Minitab and SAS. Important Note: When you use Minitab, leaving the cell blank is not a valid non-response code for the last line of data. Please use a period sign to indicate non-response in the last line of data. 7. Pre and Post Coding: Sometimes, we have enough prior knowledge to develop a complete coding scheme for a questionnaire before we collect any data. In that case, we only 200

need to do pre-coding. In other cases, we may not be able to determine all the values a variable may take beforehand. Sometimes we may not even be able to determine how many variables are there before we look at the data. Then, we need post-coding. The amount of post-coding we need to do depends on how much prior research we do. A focus group held before questionnaire construction can significantly reduce the extent of post-coding required and thereby save time needed to record the data. During coding, we should try to avoid information loss as much as possible. Examples of coding 1. Question: Residence:

Apartment Dorm Fraternity/Sorority Other Coding scheme: In this case, we are measuring only one thing: residence, that is, this question represents only one variable. Let us call this X1 . X1 is a nominal variable, and we can use any four different numbers to represent the different possibilities. A simple coding scheme will be: Apartment → X1 = 1,

Dorm → X1 = 2,

Fraternity/Sorority → X1 = 3,

Other → X1 = 4,

If non-response occurs, enter a period (.) sign. No post coding is necessary here. 2. Consider now a variation of question 1: Residence: Apartment Dorm Fraternity/Sorority Other (specify) Once again, this question involves the measurement of a single nominal variable, which we may call X2 . In this case, however, the respondent fills in what “other” means. Thus, there may be multiple categories of “other,” and we have to incorporate this fact into the coding scheme. For example, we may use the following coding scheme: Apartment → X2 = 1,

Dorm → X2 = 2,

Share house → X2 = 4, X2 = 6,

Fraternity/Sorority → X2 = 3,

Share mobile home → X2 = 5,

Home with parents →

Non-response → period sign. If we did not know all the possibilities about “other” before we studied the data, we would need to do post coding here. When we use a question like this, there may be many different responses under “other,” and some responses may be present a very small number of times. We may then create a few main categories of “other,” and lump the remainder into an “all others” category to make coding convenient. In doing this, we need to make a decision about how much information we can afford to lose in the process. 3. Question: 201

Which of the following movies have you ever seen (check all that apply): Borat The Da Vinci Code

Shrek 2 Pirates of the Caribbean 2

Cars

Coding scheme: Here, we have five yes/no questions. Thus, we need five variables; let us call them X3a , . . . , X3e . A possible coding scheme is: If “Borat” is checked, then X3a = 1. Otherwise, X3a = 0. If “Shrek 2” is checked, then X3b = 1. Otherwise, X3b = 0, etc. In this case, we do not need post coding. Also, here, we cannot tell if non-response is present since we cannot differentiate between whether the respondent is not answering a question, or just has not seen the movie. 4. Question: “In the space below, list all the movies you have seen in the last month.” Coding scheme: (i) Make a list of all the movies mentioned by the entire sample of respondents. (ii) Create a separate variable for each movie. For each movie in the list obtained from the whole sample, a given respondent is telling us whether he has seen it or not in the last month. Thus, he is effectively answering a series of yes/no questions. (iii) Example of a possible coding scheme: X4a = 1 if respondent has mentioned “Borat,” X4a = 0 if not. X4b = 1 if the respondent has mentioned “Shrek 2,” X4b = 0 if not, etc. Note: (1) Here, we need post coding unless we have prepared, beforehand, a complete list of all possible movies a respondent might see. The post coding here involves the creation of variables. (2) There will be some movies listed which very few respondents have seen (e.g., one or two respondents out of 200 have listed them). We may lump those responses into an “others” variable. (3) Here, we cannot identify if non-response is present. If no movie is listed, that may simply mean that the respondent did not see any movie during the last month. 5. Question: “(a) Do you ever listen to the radio? Yes

No (If no, please skip part (b))

(b) In the space below, list all the radio stations you listen to.”

Coding Scheme: 1. For Part (a), we are asking just one question and have one variable as a result. Let us call it X5a . We can code X5a as follows : 202

X5a = 1 if “YES,”

X5a = 0 if “No,” and period sign if non-response.

Note that if a respondent forgets to answer part (a), but lists radio stations in part (b), we can detect that fact and record X5a = 1. 2. For part (b), we need to make a list of radio stations from the entire sample, and create a separate variable for each radio station listed. (If some stations are listed by very few people, you may lump them into an “others” variable.) For each radio station listed, a respondent is effectively answering a YES/NO question. A possible coding scheme is: X5b = 1 if respondent listens to 91.3 FM, X5b = 0 if not, and period sign if non-response. X5c = 1 if respondent listens to 92.1 FM, X5b = 0 if not, and period sign if non-response, etc. Note that non response occurs here in the following two cases: (1) The respondent skips both part (a) and part (b). (2) The respondent checks “YES” in part (a), but skips part (b). 6. Question:

“Age

(years)”

It is a single variable, and usually the number provided by the respondent is entered without any change. As usual, a period sign is used for non-response. 7. Question: Example: Please rate the Wegmans grocery store on each of the following: Very Poor Excellent Don’t Know (a) Store Hours 1 2 3 4 5 6 7 DK (b) Selection of items 1 2 3 4 5 6 7 DK (c) Price 1 2 3 4 5 6 7 DK You have three variables X7a , X7b , and X7c . For each one enter the number circled, or leave the cell blank if the item is either skipped or DK is circled.

11.6

Data Editing and Cleaning

I now discuss some of the common errors in data entry and how to address them. Step 1. Check if you interchanged rows and columns: Before you do anything else, check if you mistakenly entered data as columns instead of rows, that is, entered ID followed by the variables names in Column 1 instead of Row 1, and had a column of data from each respondent. If you made this mistake, proceed to correct as follows: (1) In the EXCEL worksheet, mark the data set. In the menu line, take the cursor to Edit and click Copy. (2) Mark a cell below your data set by clicking on it. Take the cursor to Edit and click on “Paste Special.” A dialog box opens. Find “Transpose” at the bottom of the dialog box. Click on “Transpose.” This will make each column a row and vice versa. 203

(3) Delete the original data set and keep the transposed data set. Step 2. Prepare One-Way Tabulations for all Variables using Minitab: (1) In the computer screen, double click on the Minitab icon to open Minitab. (2) In the menu line at the top of the screen, click on “File” and then click on “Open Worksheet.” A dialog box opens. In the dialog box, click on the bottom line (“Files of type”) and click on EXCEL. Use the browser line at the top of the dialog box to access and open your Excel worksheet. The Excel worksheet opens as a Minitab worksheet. (3) In the menu line at the top of the screen, click on “Stat” and drag the cursor to “Tables.” Click on “Tally Individual Variables.” (4) Select all the variables in the data set by marking their names in the dialog box and clicking “Select.” Make sure at least “Count” is marked under “Display” in the dialog box. (5) Click “OK.” Minitab now displays a one-way table for each variable in the data set. Detecting Errors in Data Entry Using One-Way Tables: Examine each one-way tabulation individually. For a given variable, the table gives you the different values it takes in the sample, and the number of times it takes each value (“count”). If you find a number you do not expect to see, a error has occurred in data entry. Usually, data entry errors occur in three ways. • For a given variable, a wrong code was used. For example, suppose you coded the dichotomous variable gender as 1 for male and 0 for female. One of group members entered 2 for female by mistake. Then, in the one-way table for gender, you will see 2 as a value of the variable in addition to 0 and 1. Typically this error tends to be repeated and can be detected easily. • When entering a string of numbers, you skipped a number. For example, suppose your variables are gender (coded 1/0) and interest in watching sports on TV (1-7 scale, coded 1 to 7). If you skip gender by mistake, you will have the value of the next variable in the column for gender. Then, you may see any number from 2 to 7 in the table for gender in addition to the valid numbers 0 and 1. • Sometimes the number entered in simply wrong. When the entry is wildly wrong, it can be detected. For example, suppose the variable is how much money a student spends on cellular telephone service in a month. Suppose a respondent wrote $60 as response. If you add a zero by mistake, the data entered will be 600 instead of 60. From the one-way tables, you can detect numbers which do not look right. 204

Step 3. Identifying and Correcting Errors: Once you detected errors using one-way tabulations, you need to correct the errors. • First, examine the EXCEL worksheet and find the places where the errors occurred. Often the error is simple and can be fixed immediately. It is very helpful if you retain the original completed surveys for reference. If you can correct the entries, no data are wasted. • Sometimes, when you suspect an error, it is difficult to determine what the correct value is. This happens when the original data are not easy to access. Then, it becomes necessary to declare the wrong data entries as “missing data.” One way to do that is change the suspect cells in the EXCEL sheet to blank.

205

11.7

Example of Survey and Coding Scheme 11.7.1 Carrier Dome Events Survey

Dear Participant: We are conducting a survey to find out your opinions about the events held at the Carrier Dome. We will be grateful if you take a few minutes of your time to complete the survey. Your answers will be kept strictly confidential and will not be used for any commercial purpose. Thank you very much for your help. Students of Marketing Research, School of Management (Note: To conserve paper, the survey is printed on both sides.) Survey A. Background Information 1. Gender

Male

2. Year in School: Freshman Other

Female Sophomore

Junior

Senior

3. Where do you live? (Check one) Dorm

Fraternity/Sorority

Rental apartment/house

Other

South Campus

4. Are you a member of fraternity or sorority? Yes

No

5. How do you come to campus? (Check all that apply) Walk

Bicycle

Bus

Other

Drive

6. How far do you live from the Carrier Dome? (Check one) less than a mile 1−2 miles more than 2 miles Don’t Know 7. Which of the following are you a member of? (Check all that apply) Fraternity

Sorority

Varsity Sport

SU Athletic team

Club Sport

206

B. Recreation Habits 8. How interested are you in the following activities in your spare time? (Please circle) Not interested at all

Very interested

(a) Exercise

1

2

3

4

5

6

7

(b) Participate in sports

1

2

3

4

5

6

7

(c) Shop for clothes

1

2

3

4

5

6

7

(d) Go to bars

1

2

3

4

5

6

7

(e) Go to malls

1

2

3

4

5

6

7

(f) Watch movie

1 2

3

4

5

6

7

(g) Do volunteer work

1

2

3

4

5

6

7

(h) Study/read

1

2

3

4

5

6

7

(i) Listen to music

1

2

3

4

5

6

7

(j) Spend time with friends

1

2

3

4

5

6

7

(k) Watch sports on TV

1

2

3

4

5

6

7

(l) Watch sports at the Dome

1

2

3

4

5

6

7

207

C. Attendance of Carrier Dome Events 9. Have you ever been a season ticket holder for any of the following? (Check all that apply) SU Football SU Lacrosse SU Men’s Basketball 10. During the last one year, which of the following events have you attended at the Carrier Dome? (Check all that apply) SU Men’s Basketball game

SU Football game

SU Lacrosse game

High school sporting event

Empire State games

Concert

Commencement address

Telecast of SU Basketball game

11. During the current academic year (2002 − 2003), have you held a season ticket for any of the following? (Check all that apply) SU Football SU Lacrosse SU Men’s Basketball 12. How many people do you usually attend a Dome event with? (Check one) 0

1

2−3

4 or more

Never attend a Dome event 13. Who do you usually attend a Dome event with? (Check all that apply) Parents

Siblings

Friends

Spouse/significant other

Other

Never attend Dome event

208

14. How much money do you spend on concessions at an average Dome event? (Check one) Never attend a Dome event

$0-$5

$6-$10

$16-$20

$21 or more

$11-$15

D. Opinion of Carrier Dome 15. Please rate the Carrier Dome on each of the following by circling your response: Very Poor

Excellent Don’t Know

(a) Choice of events

1 2

3

4 5

6

7

DK

(b) Day/time of events

1

2 3

4 5

6

7

DK

(c) Quality of seating

1

2

3 4

5

6

7

DK

(d) Ticket price

1

2

3 4

5

6

7

DK

(e) Food choices

1 2

3

4 5

6

7

DK

(f) Food price

1

2

3 4

5

6

7

DK

(g) Drink choices

1

2 3

5

6

7

DK

(h) Drink price

1 2

3 4 5

6

7

DK

(i) Souvenir choice

1

3 4

5

6

7

DK

(j) Souvenir price

1 2

3

4 5

6

7

DK

(k) Cleanliness of restrooms (l) Availability of restrooms (m) Availability of parking

1

2

3 4

5

6

7

DK

1

2 3

5

6

7

DK

3

4 5

6

7

DK

(n) Safety

1

2 3

4 5

6

7

DK

(o) Overall value for the money

1

2

6

7

DK

2

1 2

4

4

3 4

5

16. Has the threat of terrorism changed your Dome attendance habits since 9/11/’01? Yes No 209

17. What should be done to improve safety at the Dome? (Check all that apply) More police presence Metal detectors at gates

Better screening at entrance

E. Interest in new Dome events 18. If a combined lacrosse/football/basketball season ticket were available, how likely would you be to purchase it at the following prices? (Please circle) Very unlikely

Very likely

(a) $50

1 2 3 4

(b) $75

1 2

(c) $100

1

(d) $125

1 2 3 4

(e) $150

1

2 3 4 5

6

7

(f) $175

1

2 3 4 5

6

7

(g) $200

1 2 3

3

5 6

7

4 5 6

7

2 3 4 5

4

6

7

5 6

7

5 6 7

19. Consider the concept: “Allow all students free access to all games at the Carrier Dome in exchange for a fixed increase in tuition fee.” How strongly do you agree/disagree with this statement? (Please circle) Strongly Disagree 1

Strongly Agree 2 3 4 5 6 7

20. In the space below, please list any event(s) you would like to see at the Carrier Dome.

**Thank You for Your Participation**

210

11.7.2 Coding Scheme and Instructions for Data Entry Dome Events Questionnaire The coding scheme for the Carrier Dome survey is provided. Please give each respondent in the data your group collected an ID starting with 1. You have 76 variables: ID, X1 , X2 , X3 , . . ., X15a − X15o , . . ., X19 . Prepare an Excel sheet. The top line of the sheet lists the variables: ID in cell A, X1 in cell B, . . ., X19 in cell BX. Then, enter a row of numbers from each respondent. For each person, ID goes to cell A, X1 to cell B, , X19 to cell BX. If there is non-response for a variable, leave the corresponding cell blank. Use the nonresponse code also when a question does not apply. For example, consider question 13. If a student never attends a Dome event, leave the five cells from question 13 blank. If a question is either skipped or does not apply, then also leave the cell blank.

211

Coding Scheme A. Background Information 1. X1 Gender

1 Male

0 Female

2. X2 Year in School: 1 Freshman 2 Sophomore 5 Other

3 Junior

4 Senior

3. X3 Where do you live? (Check one) 1 Dorm

2 Fraternity/Sorority 3 South Campus

4 Rental apartment/house 5 Other 4. X4 Are you a member of fraternity or sorority? 1 Yes

0 No

5. How do you come to campus? (Check all that apply) X5a : Walk

X5b : Bicycle

X5c : Drive

X5d : Bus X5e : Other Variables X5a − X5e : For each variable, enter 1 if checked and 0 if not. Enter five period (.) signs if question 5 is skipped. 6. X6 : How far do you live from the Carrier Dome? (Check one) 1 less than a mile 2 1−2 miles 3 more than 2 miles Period (.) sign: Don’t Know 7. Which of the following are you a member of? (Check all that apply) X7a : Fraternity

X7b : Sorority

X7c : Club Sport

X7d : Varsity Sport X7e : SU Athletic team For each variable, enter 1 if checked and 0 if not. (There is no non-response for this question.)

212

B. Recreation Habits 8. How interested are you in the following activities in your spare time? (Please circle) Not interested at all

Very interested

(a) Exercise

1

2

3

4

5

6

7

(b) Participate in sports

1

2

3

4

5

6

7

(c) Shop for clothes

1

2

3

4

5

6

7

(d) Go to bars

1

2

3

4

5

6

7

(e) Go to malls

1

2

3

4

5

6

7

(f) Watch movie

1

2

3

4

5

6

7

(g) Do volunteer work

1

2

3

4

5

6

7

(h) Study/read

1

2

3

4

5

6

7

(i) Listen to music

1

2

3

4

5

6

7

(j) Spend time with friends

1

2

3

4

5

6

7

(k) Watch sports on TV

1

2

3

4

5

6

7

(l) Watch sports at the Dome 1 2 3 4 5 6 7 Variables X8a − X8l . For each variable, enter either the number circled, or a period (.) sign if the item is skipped. C. Attendance of Carrier Dome Events 9. Have you ever been a season ticket holder for any of the following? (Check all that apply) X9a : SU Football X9b : SU Lacrosse X9c : SU Men’s Basketball For each variable, enter 1 if checked and 0 if not. (There is no non-response for question 9.)

213

10. During the last one year, which of the following events have you attended at the Carrier Dome? (Check all that apply) X10a : SU Men’s Basketball game

X10b : SU Football game

X10c : SU Lacrosse game

X10d : High school sporting event

X10e : Empire State games

X10f : Concert

X10g : Commencement address X10h : Telecast of SU Basketball game For each variable, enter 1 if checked and 0 if not. (There is no non-response for question 10.) 11. During the current academic year (2002 − 2003), have you held a season ticket for any of the following? (Check all that apply) X11a : SU Football X11b : SU Lacrosse X11c : SU Men’s Basketball For each variable, enter 1 if checked and 0 if not. (There is no non-response for question 11.) 12. X12 : How many people do you usually attend a Dome event with? (Check one) 10

21

3 2−3

4 4 or more

Period (.) sign: Never attend a Dome event 13. Who do you usually attend a Dome event with? (Check all that apply) X13a :

Parents

X13b : Siblings

X13c : Friends

X13d : Spouse/significant other X13e : Other Never attend Dome event For each variable, enter 1 if checked and 0 if not. Enter five period (.) signs if question 13 is skipped or “never attend Dome event” is checked.

214

14. X14 : How much money do you spend on concessions at an average Dome event? (Check one) Period (.) sign: Never attend a Dome event 1 $0-$5 2 $6-$10 4 $16-$20 5 $21 or more D. Opinion of Carrier Dome

3 $11-$15

15. Please rate the Carrier Dome on each of the following by circling your response: Very Poor

Excellent Don’t Know

(a) Choice of events

1 2

3

4 5

6

7

DK

(b) Day/time of events

1 2 3

4 5

6

7

DK

(c) Quality of seating

1

2 3 4

5

6

7

DK

(d) Ticket price

1

2

5

6

7

DK

(e) Food choices

1 2

3

4 5

6

7

DK

(f) Food price

1

2 3

4 5

6

7

DK

(g) Drink choices

1 2

4

5

6

7

DK

(h) Drink price

1

2 3

4 5

6

7

DK

(i) Souvenir choice

1

2 3

4 5

6

7

DK

(j) Souvenir price

1

2

6

7

DK

(k) Cleanliness of restrooms (l) Availability of restrooms (m) Availability of parking

1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7

DK DK DK

(n) Safety

1

DK

3 4

3

3 4

2 3

5

4 5

6

7

(o) Overall value 1 2 3 4 5 6 7 DK for the money Variables X15a − X15o : For each variable, enter either the number circled, or a period sign if the item is skipped or DK is circled. 16. X16 : Has the threat of terrorism changed your Dome attendance habits since 9/11/’01? 1 Yes 0 No 215

17. What should be done to improve safety at the Dome? (Check all that apply) X17a : More police presence X17c : Metal detectors at gates

X17b : Better screening at entrance

For each variable, enter 1 if checked and 0 if not. (There is no non-response for question 17.) E. Interest in new Dome events 18. If a combined lacrosse/football/basketball season ticket were available, how likely would you be to purchase it at the following prices? (Please circle) Very unlikely

Very likely

(a) $50

1 2 3 4

(b) $75

1 2

(c) $100

1

(d) $125

1 2 3 4

(e) $150

1

2 3 4 5

6

7

(f) $175

1

2 3 4 5

6

7

3

5 6

7

4 5 6

7

2 3 4 5

6

7

5 6

7

(g) $200 1 2 3 4 5 6 7 Variables X18a − X18g : For each variable, enter either the number circled, or a period (.) sign if the item is skipped. 19. X19 : Consider the concept: “Allow all students free access to all games at the Carrier Dome in exchange for a fixed increase in tuition fee.” How strongly do you agree/disagree with this statement? (Please circle) Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 Enter either the number circled, or a period (.) sign if the item is skipped. 20. In the space below, please list any event(s) you would like to see at the Carrier Dome. Coding: Do not enter anything from question 20 into the Excel sheet. On a separate page list the new events and record how many times each event is recommended. **Thank You for Your Participation**

216