Crm Analytics.pdf

BEP386

CHAPTER SEVEN

Customer Relationship Management Analytics From Relationship Marketing Re-Imagined:

Marketing’s Inevitable Shift from Exchanges to Value Cocreating Relationships By Naresh K. Malhotra, Can Uslay, and Ahmet Bayraktar (A Business Expert Press Book)

Copyright © Business Expert Press, LLC, 2016. All rights reserved. Harvard Business Publishing distributes in digital form the individual chapters from a wide selection of books on business from publishers including Harvard Business Press and numerous other companies. To order copies or request permission to reproduce materials, call 1-800-545-7685 or go to http://www.hbsp.harvard.edu. No part of this publication may be reproduced, stored in a retrieval system, used in a spreadsheet, or transmitted in any form or by any means – electronic, mechanical, photocopying, recording, or otherwise – without the permission of Harvard Business Publishing, which is an affiliate of Harvard Business School.

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

CHAPTER 7

Customer Relationship Management Analytics Introduction This chapter introduces the role and benefits of customer r­elationship management (CRM) and the historical development of the CRM ­concept. In addition, it discusses how to measure and manage customer value and customer lifetime value (CLV) by presenting various approaches and principles. Besides, it introduces two types of strategies, namely, “across-customer” strategies and “within-customer” strategies, which help companies maximize CLV. The chapter also discusses the importance of CLV-based marketing initiatives, which channel firms’ efforts toward customer-centric marketing and away from product orientation. After introducing CRM solution providers, the chapter concludes with two cases that illustrate how CRM technologies can help firms grow their business and play a key role in achieving long-term success by transforming the whole company.

Are Loyal Customers Profitable Customers? Behavior loyalty is not a very good predictor of retention and represents an even worse predictor of customer satisfaction. For example, customers under service contracts can appear to be loyal for the duration of the contract but switch to a competitor with better pricing as soon as the contract expires. Satisfaction is a matter of meeting or exceeding past expectations; however, loyalty also includes an expectation of future actions.1 Therefore, the two do not always go hand in hand. According to one study, 75 percent of consumer wireless phone service customers were found to be satisfied, but 72 percent would be willing to switch to a competitor.2

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

104 RELATIONSHIP MARKETING RE-IMAGINED

Even attitudinal loyalty may not necessarily reflect in profitability. A customer may be really happy with a BMW that he or she purchased at a year-end sale and have very favorable expressed loyalty toward the brand. However, that becomes much less meaningful to the company if the customer enjoys the car for some 20 years and never gets it serviced at a BMW dealer once the warranty expires. These nuances underline why CRM applications have become increasingly important during the past three decades.

Effective CRM Strategy and Measurement

The former Union Trust Bank in San Francisco, now Wells Fargo. Firms such as Wells-Fargo Bank have invested millions of dollars for hardware as well as employee training to effectively manage their customer relationships in order to maximize their profits. Source: Wikimedia by Kjteil Ree.

The CRM process begins with customer selection (based on the mission statement of the firm), developing the value proposition to serve them, monitoring nonfinancial metrics of performance, and finally making sure that these metrics are ultimately reflected in financial performance, that is, customer profitability. However, in many cases, switching from mass to relationship marketing may also require a transformation of the culture of the firm from the top all the way to the bottom. The technical aspects of CRM, although expensive, are relatively easy to handle. It is the human

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 105

aspects of the CRM transformation that is challenging and time consuming in developed as well as developing nations.3 The primary advantages of CRM over mass and direct marketing include the ability to increase customer satisfaction, effective reach, automation, contextual individualization, and experience-based branding. Its disadvantages include high organizational involvement and costs.4 Once properly implemented, the benefits of CRM are numerous: • CRM enables the marketer to measure and manage to optimize the cost to serve customers. Metrics can be tracked at the individual level. • It empowers the firm to determine which customers are not profitable and “outsource” them if necessary. There is no need to serve a persistently unprofitable customer with poor strategic prospects. • CRM marketers can conduct marketing experiments by sampling from different categories and measure impact across groups. This allows for fine-tuning of marketing campaigns at relatively low cost. • CRM enables the measurement of return on marketing investment (ROMI). Campaigns enabled by CRM are more target-oriented and have a better impact. • Thus, CRM increases customer retention as well as customer acquisition. • CRM is a dynamic learning system and becomes more effective with more data.5 In order to serve the customers effectively, it is important to recognize that customer relationships are dynamic and evolve over time. The relationships can be functional, structural, and strategic in nature and marketers need to actively pursue customers to build relationships with them. Point to Ponder: Consider the following definition of CRM from The Wall Street Journal: Customer Relationship Management (CRM): The process of storing and analyzing the vast amounts of data produced by

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

106 RELATIONSHIP MARKETING RE-IMAGINED

sales calls, customer-service centers and actual purchases, supposedly yielding greater insight into customer behavior. CRM also allows businesses to treat different types of customers differently—in some cases, for instance, by responding more slowly to those who spend less or charging more to those who require more extensive handholding.6 What do you think of this definition?

Historical Development of CRM Although relationship marketing and CRM are often used interchangeably, CRM typically refers to the technology-based solution domain of relationship marketing, often confused with CRM technology by itself.7 It has been used to refer to a wide array of applications, including sales force automation, direct mail campaigns, loyalty programs or databases, and help desk or call center. To some it was about populating a data warehouse or undertaking data mining; others considered CRM an e-commerce solution, such as the use of a personalization engine on the Internet or a relational database for Sales Force Automation. This lack of a widely accepted and appropriate definition of CRM can contribute to the failure of a CRM project when an organization views CRM from a limited technology perspective or undertakes CRM on a fragmented basis.8 To add to the confusion, at least a dozen definitions of CRM have been proposed. We adopt and advocate the following definition: “CRM is a comprehensive strategy and process of acquiring, retaining, and partnering with selective customers to create superior value for the company and the customer.”9 Additional features of ideal CRM are that it embraces e-commerce, disseminates through the whole organization, promotes and requires long-term vision, enables customization, and is data driven, information intensive, and outcome oriented. Having said all this, the following three main assumptions about long-life (behaviorally loyal) customers have been tested but found to be

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 107

unfounded: (a) the costs of serving long-life customers are less, (b) they pay higher price premiums, and (c) they spend more. Therefore, behaviorally loyal customers are not necessarily more profitable than other customers.10 Firms need to be able to differentiate their customers by their (current and potential, long term) value and implement CRM initiatives that are aligned with providing the customer better value. Small volume customers can be profitably reached via direct marketing campaigns. A well-crafted direct marketing campaign is not wasteful but can be very effective. For example, Dell considers their direct marketing skills as a key competitive advantage, more so than their products and manufacturing.11 Customer value is not built one transaction at a time but rather through a series of mutually successful episodes. Some episodes will inevitably include failures. The way the firm handles these failures and recovers from them can create even more loyal and devoted customers. Yet, roughly half of the customers with complaints are not happy with the way companies handle their complaints.12 In the airline industry, it has been shown that customer complaints can destroy firm value more than satisfaction contributes to it.13

Point to Ponder: Why do you think that CRM strategies are successful? Point to Ponder: What is the difference between brand management and CRM?

Measuring and Managing Customer Value For many firms, the 80:20 rule has become even more skewed and only a few of their customers may generate most of their revenues and almost all of their profits. For such firms, management of key accounts is vital. Each relationship must be cultivated and developed, and hopefully cemented over time. Alas, even the firms who have long segmented their customers into value categories have not been doing a good job of it. For example, with the exception of the top 10 percent of fliers, the assigned category tiers do not correlate with the customers’ true values in the frequent-flier databases.14 Poor implementation and lack of appropriate technology are the two main causes of this recurring problem.

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

108 RELATIONSHIP MARKETING RE-IMAGINED

Various metrics are available for measuring the value of the customer and implementing the CRM strategies. Some traditionally used metrics are recency–frequency–monetary (RFM) value, share of wallet (SOW), and past customer value (PCV). These existing metrics have several limitations and do not link the value of a customer to future profitability. On the other hand, CLV is an advanced metric that is widely gaining popularity across all sectors and industries. The forward-looking CLV metric considers the future value of a customer to the firm and aids in designing and implementing marketing strategies for the present, thus maximizing profitability.15

RFM Approach On the basis of the assumption that past purchase behavior of customers better predicts their future purchases than demographic data, RFM scoring models have been used in marketing for more than four decades. Historically, the initial applications were in business-to-consumer (B2C) contexts: direct marketing, insurance, banking, telecommunications, and so on. The customers are ranked based on their R, F, and M score. Recent purchasers, frequent purchasers and big spenders, on average, receive higher scores. The higher the score is, the more profitable is the customer, which is expected to continue in the future. A basic ranking, within parent cell ranking, or weighed cell ranking can be used. In basic RFM, the customers are scored on each factor separately. That is, the customers are ranked based on the recency of their purchases and then split into subgroups (e.g., quintiles). As a result, there could be five quintiles where the top quintile represents the 20 percent that has made the most recent purchases. The procedure is then repeated for frequency and total money spent and reveals 5 × 5 × 5 = 125 cells from highest to lowest ranking. Within parent cell ranking, the cells are grouped based on recency, similar to the basic approach. However, each group is then subgrouped based on frequency (rather than separately). Each of the resulting 25 cells is finally ranked by monetary value. This approach requires more sorting than the former. In weighed cell ranking, weights are assigned to the metrics and sorted accordingly. These weights can be equal; however, an often used weight scheme is: (3 × R) + (2 × F) + (1 × M). The weights used depend on the This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 109

data and specific application. For example, based on the response rate, M may turn out to be the most important factor in some cases.16

Share of Wallet SOW refers to the share of a company of a customer’s expenditure for a given offering category. Although market share has been shown to uniquely impact the financial performance of the firms based on their strategy types,17 and market share and SOW are linked, targeted efforts to increase the share of large wallet customers can have more direct and positive bottom line impact. For example, major casinos “comp” their high rollers (or “whales”) with the hope to attract them to gamble in their facilities whenever they visit, that is, they hope to attract close to 100 percent share of gambling wallet. In other cases, firms target much smaller shares. A local restaurant would be happy to consume $50 a week of the dining wallet of a customer who spends $600 a month ($50 × 4 weeks/$600 = roughly 33% share of dining wallet). The data is typically acquired through primary surveys, which is then combined with other predictors of behavior in a database (behavioral and demographic variables) and share-of-wallet is then modeled as the dependent variable. Sj Share-of-Wallet: J ∑ j =1 Sj where S is the actual sales to the focal customer, J refers to firm in category, and the denominator represents the value of sales made by all firms to the focal customer. Point to Ponder: Are there cases where it is possible to conduct SOW analysis without having conducted survey research?

Past Customer Value The simple logic behind the PCV approach is that history is the best predictor of future. The model uses extrapolation on the previous transactions of a customer (i.e., past profits) to predict the customer’s future value. Time value of money is used to adjust previous contributions and calculate PCV for each customer. It can be calculated as This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

110 RELATIONSHIP MARKETING RE-IMAGINED

PCV = ∑ t =1GCit * (1 + d ) T

t

where i is the customer, d is the applicable discount rate, T is the number of periods prior to the current period, and GCit is the gross contribution of the transaction of customer i in time t. For example, let’s assume that a customer spent $60 (last month), and $100, $110, $190, and $210 (in preceding months) on his or her cable TV bundle package. Assuming an average gross contribution margin of 50 percent and a discount rate of 12 percent per annum, we can calculate the PCV of the customer as follows: GC = 50% × spending amount PCV Score = $30 × (1 + 0.01) + $50 × (1 + 0.01)2 + 55 × (1 + 0.01)3 + 95 × (1 + 0.01)4 + 105 × (1 + 0.01)5 = $343.19 Repeating the exercise for all the customers in the portfolio would enable prioritization of all customers and allocation of marketing resources based on their values.

Comparing the Traditional Measures to CLV As mentioned in the beginning of this section, RFM is a relatively simple, traditional technique that is relatively low cost. Its data requirements are relatively low, and it provides positive return on investment (ROI). However, it is important to provide an understanding of more sophisticated models, because RFM is currently considered to be an outdated technique. For example, it has been shown that CLV approach is generally superior to RFM. The top 5 percent of CLV rankings provide 1.6 times the net profits than the top 5 percent of RFM rankings, and profitability improves by as much as 67 percent.18 Nevertheless, recency of purchases has been shown to be linked to CLV in a nonlinear S-shaped form.19 A drawback of the SOW approach is that it may put too much ­emphasis on maximizing SOW, whereas this may not be the best long term objective for both the business and the customer. Excessive SOW and communications to maximize it may induce retaliatory variety-­seeking behavior This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 111

in B2C or antitrust concerns in business-to-business (B2B) c­ontexts. ­Furthermore, PCV and other techniques do not take expected customer activity into account. They simply extrapolate based on previous activity and do not consider future account maintenance costs. Point to Ponder: Can you provide a ranking of the effectiveness of the metrics discussed so far? What are the pros and cons of each?

A Primer on CLV A main advantage of CLV approach is that it incorporates not only the future expected value of a customer but also the probability of that ­customer remaining active in future. In contrast with the former methods, CLV enables optimal allocation of resources at the individual ­customer level. Segment-Wise Calculation of CLV In the simplest sense, if we assign (assume) a value to the revenue generated by a customer and the cost of serving that customer, we can derive the margin that the customer contributes to the bottom line of the firm in a given time frame (typically a year). CLV = ∑ t =1 T

where

( pt − ct ) rt (1 + d )t

− AC

T is the time horizon for the calculations pt is the price paid by the customer at time t ct is the cost of serving the customer at time t rt is the probability of customer repeat purchase at time t d is the discount rate or cost of capital for the firm AC is the customer acquisition cost Let’s consider the case of a subscription magazine targeting college students. Assume that the average subscription lasts four years, annual margin per subscriber is $12, annual retention rate is 90 percent, subscriber acquisition cost is $20, and the interest rate is 5 percent. Using This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

112 RELATIONSHIP MARKETING RE-IMAGINED

the above formula, estimated lifetime value (LTV) of a subscriber would be merely $18.30. However, let’s assume that the magazine has a loyal following that extends beyond college years. In this case, we can assume that the margin (M) is fixed over time, and simplify the LTV formula by assuming that the life cycle is infinite (i.e., N  ∞). Thus, the basic CLV equation for infinite economic life becomes ∞

mr t

t =0

(1 + d )

CLV = ∑

t

=m

r − AC 1− r + d

In this scenario, the LTV per subscriber of the magazine is significantly higher, $72 − $20 = $52. Assuming an infinite lifetime cycle and constant retention rate, we can compute margin multiples (see Table 7.1). For example, with 95 percent retention and 5 percent discount rate, CLV is 9.5 times the margin per customer, whereas with 65 percent retention and 12.5 percent discount rate, it is only 1.37 times the margin. It has been estimated that a 5 percent increase in customer retention improves profitability by 25 to 100 percent! Given the prominence of r in the above equation, it is not hard to imagine how this dramatic effect comes about. Some (but not all) retained customers may pay price premiums, buy more, generate positive word of mouth, and enable cost savings (may cost less to serve and decrease overall acquisition costs).20 For example, Harrah reported that its loyalty program improved its share of gambling wallet from 36 to 53 percent; it also increased the slot revenue by 12 percent by using its CRM database to redesign the casino floor. In the process, it also increased its occupancy rates and revenue from top-tier guests.21 Table 7.1 The margin multiple Retention rate (r)

Discount rate (i) 5%

7.5%

10%

12.5%

65%

1.6

1.53

1.44

1.37

75%

2.50

2.31

2.15

2.00

85%

4.25

3.77

3.4

3.09

95%

9.5

7.6

6.34

5.42

Source: Adapted from Gupta and Lehmann (2008).

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 113

The margin (m) has been found to be relatively stable over time.22 However, we can also incorporate a growth rate to the equation (assuming 1 + i is larger than r(1 + g)): CLV = m

r 1 + i − r (1 + g )

Instead of an infinite life, the expected lifetime of the customer (i.e., 1/(1 − r)) may also be used for estimating CLV. For example, with 90 percent retention rate, the expected lifetime is 10 years. Although simple and straightforward, it has been shown that the expected lifetime approach overestimates CLV.23

Advanced Measurement of CLV The CLV models presented above make several simplifying assumptions. For example, the retention rate is constant and applies equally to the customer groups. However, in many cases, retention probability changes drastically from one customer (segment) to another customer (segment) and should not be averaged across customers. Generally speaking, making the models more realistic means making them more complex. For ­example, when we account for returns on acquisition, retention, and add-on selling, we observe24 CE(t ) =

I



∞



k



i =0



k =1



j =1



∑  N i ,t ai ,t (Si ,t − ci ,t ) − N i ,t Bi ,a,t + ∑N i ,t ai ,t  ∏ ri ,t + k 

(

k  1   Si ,t + k − ci ,t + k − Bi ,r ,t + k − Bi , AO , t + k    1 + d   

)

where CE(t) is the customer equity value for customer acquired at time t Ni,t is the number of potential customers at time t for segment i ai,t is the acquisition probability at time t for segment i ri,t is the retention probability at time t for a customer in segment i Bi,a,t is the marketing cost per prospect (N) for acquiring customer at time t, segment i

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

114 RELATIONSHIP MARKETING RE-IMAGINED

Bi,r,t is the marketing in time period t for retained customers for ­segment i Bi,AO,t is the marketing costs in time period t for add-on selling for segment i d is the discount rate Si,t is the sales of the product or services offered by the firm at time t for segment i Ci,t is the cost of goods at time t for segment i I is the number of segments J is the segment designation t0 is the initial time period k is future time period counter The average CLV can then be calculated by dividing CE by the total number of customers. Average CLV metric is useful to evaluate the competitors (good estimates of competitor CLVs can be calculated based on publicly available data) and can also be used as a proxy of a firm’s market valuation for high growth companies. However, it does not render itself for one-to-one marketing approaches at the individual level.25 Point to Ponder: The Thomas family, an older empty nest couple, who normally take domestic trips for vacation, decided to fly around the world on a special promotion and spent $6,900 on fares. On the other hand, the Arjona family, a middle-aged couple with no kids, usually takes one international trip to their home country a year and spends $900 on airfares. If you are a travel agent serving both families, should the Thomas or Arjona family have priority in your marketing efforts for the next five years? Which effort would generate more immediate returns? Explain. (Assume that the domestic travel profile of the families is equivalent.)

Individual CLV Calculation The calculation of CLV includes determining the (time-adjusted) future contribution margin, future costs, and customer retention. In the following section, we discuss the components and the method of calculating contribution margin and future costs. Expressed in general form, CLV is

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 115

CLVi = ∑ t =1 T

Future Contribution Margin it − Future Cost it

(1 + d )t

where i is the customer index, t is the time index, T is the forecast horizon d is the discount rate In contrast with contractual setting where prediction of the customer retention metric is emphasized, the focus is on predicting future customer behavior and contribution margin in noncontractual settings (where customer defection rates are significantly higher).26 P(Active) refers to the probability that a customer continues to purchase from the firm in the future. This is calculated at the unique individual level in contrast with the aggregate retention rate. The net present value (NPV) of expected gross contribution (EGC) can be calculated as27 t +x

NPV of EGCit = ∑ n =t +1 P ( Active)in ×

AMGCit

(1 + d )n

where AMGCit is the average gross contribution margin in period t based on all prior purchases i is the customer index t is the period for which NPV is being estimated n is the number of periods beyond t d is the discount rate P(Active)in is the probability that customer i is active in period n As an example, let’s use the example of a customer who spends $60, $100, $110, $190, and most recently, $210 on his or her cable bundle package, respectively. Assuming an average gross contribution margin of 50 percent, a discount rate of 12 percent per annum, and the probability

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

116 RELATIONSHIP MARKETING RE-IMAGINED

of the customer remaining active is 50 percent for the next month and 35 percent for the month after that, we can calculate the NPV of EGC as follows: AMGC = (60 + 100 + 110 + 190 + 210)/5 = $134 NPV of EGC = 0.5 ×

134

(1 + 0.01)

1

+ 0.3

134

(1 + 0.01)2

= $105.7445

P(Active) is calculated based on frequency and time between purchases. That is, T  P ( Active ) =   N

n

where n is the number of purchases made during the observation period T is the time between first and last purchase N is the time between first and focal period for prediction For example, consider the following purchase schedule by three customers (Table 7.2). If we want to estimate P(Active) for week 10, 2

5 P(Active)Ali =   = 0.25  10  2

5 P(Active)Ben =   = 0.31  9 5

6 P(Active)Charlize =   = 0.08  10  Table 7.2  Purchase frequency of three customers Customers Week Week Week Week Week Week Week Week Week Week 1 2 3 4 5 6 7 8 9 10 Ali

1

Ben Charlize

1 1

1

1

1 1

1

1

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 117

On the basis of previous purchase frequency patterns of Ali and Ben, it would not be unlikely for them to make a purchase during week 10. However, Charlize, who was previously the most active, has not purchased in a while and is not likely to purchase in the future. Ali is about three times more likely to remain an active customer than Charlize. The basic equation to calculate P(Active) can certainly be improved and customized to serve a particular context better. Marketing cost usually leads the market response (sales and profits). Therefore, assuming that marketing costs take place in the beginning of a period and that gross contribution is observed at the end of a period, CLV formula can be stated as CLVi =

t +x

AMGC

x

 1 

∑ P ( Active)in × (1 + d )nit − ∑Min ×  1 + d  n = t +1 n =1

n −1

− AC

Assumptions of the model include that when a customer defects, he or she does not return (customers who return after defection are treated as a new customer by the model). This is also called as “lost for good” scenario. Naturally, this is not necessarily true in the real world, and the model underestimates CLV. To incorporate the possibility of a customer coming back and enable a second LTV, predicted frequency can be included in the CLV model:28 CLVi =

Ti

CMi , y

∑ (1 + r ) y y =1

frequency i

n

−∑

l =1

∑ mci ,m,l × xi ,m,l (1 + d )l −1

where CLVi is the LTV of customer i CMi,y is the predicted contribution margin from customer i in purchase occasion y d is the discount rate ci,m,l is the unit marketing cost for customer i in channel m in year l xi,m,l is the number of contacts to customer i in channel m in year l ­frequencyi is the predicted purchase frequency for customer i n is the number of years to forecast Ti is the predicted number of purchases by customer i until the end of planning period

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

118 RELATIONSHIP MARKETING RE-IMAGINED

Rust, Zeithaml, and Lemon CLV Model This model adopts the notion that a portion of the dormant customers can return and make purchases.29 The utility model incorporates for both CLV drivers and brand inertia. Uijk = b0k LASTijk + Xik b1k + ε ijk where b0k is the logit regression coefficient for inertia b1k is the column vector of logit regression coefficients corresponding to the drivers ε ijk is the random error term with double exponential distribution The individual choice probability for customer i is modeled as Pijk* = Pr (customer i chooses brand k*, given that brand j was chosen most recently) = exp(U ijk ∗ ) ∑ k exp(U ijk ) Customer switching is modeled based on Markov J × J matrix where J is the number of brands, and switching probabilities is pijk, where customer i having recently purchases brand j purchases brand k on next purchase. Tij

−t

CLVij = ∑(1 + d j )

f ivijt pijt Bijt

t =0

where Tij is the number of purchases customer i is expected to make before firm j’s time horizon Bijt is a firm specific element of Bit So CEj = meani(CLVij) × total number of customers in the market and CES j =

CE j

∑ kCEk

Additional advanced models for customer retention are presented in Table 7.3.

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 119

Table 7.3  Models of customer retention Models of customer retention­ Logit or probit models

Hazard models with accelerated failure time (AFT) Generalized gamma hazard model

Proportional hazard models

Pareto/NBD probability model29

Representation

P (Churn ) =

Remarks

X represents the covariates; commonly used in industry because of simplicity

1 1 + exp( β X )

t is the purchase duration for customer j and X represents the covariates

ln t j = β j X j + σµ j

( )

f (t jk ) =

g Γ(a )l

 t jj  −  a g −1  l j  t e a g jk

λ (t ; X ) = λ0 (t ) exp( β X )

g

α and γg are shape parameters and λlj is the scale parameter for customer j. lj varies across customers with an inverse generalized gamma distribution l is the hazard rate specified as a function of the baseline hazard rate l0 and covariates (X)

P ( alive|r , a , s , b , X = x , t ,T )

R and α gamma distribution parameters for customer heterogeneity r + x s   a + T   b + T  s in transactions; s and = 1 +     b for dropout rates; x  r + x + s  a + t   a + t  is the number of past s transactions of the  b +T  F ( a1 , b1 ; c1 , z1 t ) −  customer; t is the time   a +t  since trial and the most recent transaction; T is −1 time since trial and F is   F ( a1 , b1 ; c1 , z1 T   the Gauss hypergeomet  ric function

()

( )

Markov’s model

t

T −1 V ′ = ∑ t = 0 (1 + i ) P  R  

V′ is the vector of CLV, P is the transition probability matrix, R is the margin vector (both assumed constant over time)

Source: Adapted from Gupta and Lehmann (2008).

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

120 RELATIONSHIP MARKETING RE-IMAGINED

Point to Ponder: Which model has the most realistic assumptions? How are they different from one another?

Strategies for Maximizing CLV In order to maximize CLV, companies can follow two types of strategies: “across-customer” and “within-customer”. Across-customer strategies involve (a) efficient customer selection by targeting customers with high profit potential, (b) managing existing set of customers and rewarding them based on their profit potential, and (c) monitoring customer behavior and making timely interventions to prevent attrition and thereby ensuring future profitability. “Within-customer” strategies aim at maximizing profits by increasing the revenue or reducing the cost or by doing both. The “within-customer” strategies are promoting multichannel shopping (revenue maximization), optimizing allocation of resources (cost reduction), and managing the purchase sequence of the customers (revenue maximization and cost reduction). Maximizing the brand value is another key “within-customer” strategy. Customer Selection As mentioned previously, all loyal customers are not necessarily profitable. It is important to identify which customers have profit potential and focus on attracting and retaining those customers. The use of CLV approach has been proven to be superior to other traditional methods presented earlier for both B2B and B2C contexts.30 Managing Existing Customers It is also appropriate to segment customers based on CLV and develop proper marketing strategy to improve the ROI for each segment. They can then be combined with additional demographic and other variables to understand the reasoning behind each segment (see Figure 7.1). Different marketing methods maybe more suitable for different segments from both cost and response point of view. For example, reaching the

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 121

Drivers of value: • Customer lifetime • Purchase frequency • Purchase size • Price points • Cross-buying • Customer income • Marketing activities of firm • Loyalty program • Direct marketing • Network effects (e.g., word-of-mouth)

+

Customer lifetime value (CLV)

–

Detracters of value: • Cost of customer acquisition • Cost of customer retention • Cost of returns • Cost of marketing activities • Discount rate • Returns

Figure 7.1  Factors impacting customer lifetime value Source: Adapted from Singh and Jain (2009, 39).

Profitability

Butterflies True friends • Market for more profit with each • Market and invest for best profit High to create attitudinal and transaction behavioral loyalty

Strangers

Barnacles

• Ensure profit on each Low transaction; if not outsource

Short-term

• Improve profitability by investing for share of wallet when wallet is large

Customers

Long-term

Figure 7.2  Segmentation based on customer lifetime profitability and relationship duration Source: Adapted from Reinartz and Kumar (2002, 93).

customer via phone call for high-CLV customers and e-mail contact to low-CLV customers may be more justified. However, other choices may not be as straightforward, for example, a middle-CLV category may by-and-large prefer e-mail over direct mail because of their Internet competency. This approach can also be used to develop customer profiles. For example, ­Figure 7.2 categorizes customers into four groups based on

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

122 RELATIONSHIP MARKETING RE-IMAGINED

their profitability and relationship duration. There is a need to convert “butterflies” and “barnacles” to “true friends” and make profit with each purchase of the “strangers.” Monitoring Customers to Allocate Resources Optimally Some advanced models even allow for heterogeneity at the customer level. For example, one particular high-CLV customer (whale) may refuse to be contacted by phone at all times and only visit Las Vegas during a certain season and so on. Recent developments in modeling have enabled ­marketers to estimate optimal marketing spending at the individual level. There is a need to estimate the interpurchase duration of each customer, cash flows from each customer, and then maximize profits accordingly. Then the firm would know what the optimal time interval to contact a customer is, as well as the best mode of communication with him or her. More contact is not only costly but also not always better and can create tedium. Given the tremendous amount of data collected by retailers, it is possible to correlate a customer’s purchases with those who have bought the same or similar products and predict what he or she might buy next. The more relevant the communication is, the higher is the likelihood of the customer to respond to the communication with a purchase. Therefore, the sequence of purchases, timing, and value of the next purchase needs to be estimated.31 Marketing Mix Strategies As we emphasized during the introduction, value cocreation must be the focus of activities. First, for products or services, the customers should be engaged in the design, as well as the production and consumption of the service. When this happens, mass customization should improve the CLV dramatically. For example, NikeID Web site empowers the customers to design and personalize their own athletic shoes. Second, value-added services help differentiate core offerings and improve CLV.32 As for pricing, these can be adjusted based on CLV so that low-CLV customers pay more for each transaction and others are rewarded based on

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 123

their strategic importance. Multiple channels can be used to serve highCLV customers to provide added convenience. Customers served through multiple channels tend to buy more. Introducing human touch to highCLV customers can also improve their commitment to the relationship. Promotional efforts must be interactive and focus on both behavioral and attitudinal loyalty. Value cocreation can extend to the whole spectrum: coconception (military and defense contracts), codesign (Boeing and United Airlines), coproduction (Ikea), copromotion (word of mouth), copricing (eBay, negotiated pricing), codistribution (magazines), coconsumption (utility), comaintenance (patient–doctor), codisposal (self-serve), and even co-outsourcing (captive business process outsourcing). Networks that marketing interacts with to connect structural gaps include consumer, distributor, ­supplier, regulatory, and competitor networks. With this broadened ­perspective, cocreation is likely to result in an aggregate optimal value that is greater than the sum of two (or more) local optima, as in the case of exchange.33 Point to Ponder: What strategies do you think may be relevant for an industrial manufacturing firm?

Link Between CLV and Customer Equity to Shareholder Value The links between CLV and shareholder value become more obvious once the focus of the marketing manager shifts from expense and revenue to investment and assets.34 Customers respond to marketing activities through their mindset metrics such as satisfaction and attitudes, which in turn boost firm value. Marketing actions begin with tactical actions such as advertising and service improvements, new product and channel introductions, and so on. Subsequently, these marketing actions make an impact on customers, improving their attitudes and satisfaction. These improvements are expected to lead to higher brand

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

124 RELATIONSHIP MARKETING RE-IMAGINED

and customer equity and increase sales, market share, cash flow, ROI, and ultimately the value of the firm as measured by market capitalization (shareholder value).35

Implementing CLV Implementation of CLV to B2C contexts is harder, as gathering the necessary data from all customers may be cost-prohibitive and time-­ consuming. Furthermore, the firms that rely on channel intermediaries may not be in direct contact with their customers. In fact, developments in CRM technology and availability of computing power provide opportunities (through sampling if not for the entire customer base) for large as well as small firms for both B2B and B2C contexts. Therefore, the context to consider and plan for becomes B2B2C. The challenges of ­legacy CRM systems and the rationale to upgrade them include their high maintenance costs, minimal flexibility, limited functionality, poor integration, and minimal analytical capabilities. For example, one European carrier reportedly spends $2M to manage its 500,000 members using a legacy system, that is, $4 per member every year spent just to maintain the ­system. Moreover, it takes six months for the carrier to simply change a tier requirement. The same feat takes only a day with contemporary ­systems. Nevertheless, a primary reason for not successfully implementing CRM is viewing it as solely a ­technology initiative.36

Shift from Product-Centric to Customer-Centric Framework We anticipate that the use of relationship marketing will become even more prominent over time. Corporations will have CLV-based marketing initiatives that channel their efforts toward customer-centric marketing and away from product orientation.37 The producers will need to get to know their end users, and such learning will reveal new horizons for value cocreation.38 Prioritization among customers based on expected value is a worthy goal for marketers. Interestingly, prioritization not only significantly improves the relationships (hence the sales and profitability) with top-tier customers but also does not negatively affect those with

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 125

bottom-tier customers. Prioritization also decreases marketing and sales and costs, thereby increases the efficiency of the organization.39 We have discussed the many reasons why firms should engage in relationship marketing. However, the approach would not work unless the customers find value in it as well. There are numerous personal, sociological, and institutional factors that make relationship marketing appealing for the customers. These include achieving greater efficiency in decision making, simplifying the task of information processing, achieving more cognitive consistency, reducing perceived risks associated with future choices, adhering to norms of behavior set by family members, and influence of peer groups (i.e., social class and reference groups), government mandates, religious tenets, employer influences, and marketer-induced policies.40 Point to ponder: Can you think of cases where the customer does not want to build a relationship with the marketers or brand? What are the societal implications of abandoning relationship marketing in these cases? Should building relationships be-all and end-all of marketing?

CRM Solution Providers Companies such as Oracle, PeopleSoft, SAP, e.Piphany, Chordiant, NCR Teradata, Broadvision, Salesforce.com, and Kana provide CRM suites. However, these applications mostly cater to large enterprises and B2B applications. Largest consulting firms (McKinsey, Bain [corporate ­strategy]; Peppers & Rogers, Vectia [CRM strategy]; Accenture [organizational design]; Unisys, Siemens [infrastructure and systems integration]) are specialized and tend to handle the installations of these solutions, which tend to be very costly. Some vendors such as IBM, Oracle, and Sun enable build-it-yourself solutions. Others (e.g., EDS, CSC, Acxiom) specialize in infrastructure and business process outsourcing. However, despite their claim to be “complete CRM solution providers,” few software vendors can provide the full range of functionality that a complete CRM business strategy requires.41 Consultants and smaller vendors tend to serve small firms and B2C application needs.

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

126 RELATIONSHIP MARKETING RE-IMAGINED

Point to Ponder: Do you think the state of B2B or B2C CRM is more advanced? Why?

First issue of the Reader’s Digest, February 1922. The Trusted Media Brands’ subscriber database will still be the key to long-term success of the company. Source: Wikimedia.

Case-in-Point: Trusted Media Brands or Reader’s Digest Association Trusted Media Brands, formerly known as Reader’s Digest Association (RDA), first launched in 1922, has been successfully using direct mail to establish relationships with its domestic and international audiences This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 127

for almost a century. Over time, its magazine subscriber database has become a valuable asset for the company. When the company segmented its ­subscribers by level of usage, it found that, the more products a customer has bought, the higher is the likelihood of further purchases. In this case, the length and breadth of the relationship predicted future sales much better than traditional sociodemographic segmentation variables. Today, Trusted Media Brands, Inc. (TMBI) has offices in 42 countries and sells books, magazines, music, video, and educational products to a customer database of more than 140 million in 76 countries through direct marketing. It publishes 75 magazines (nine with more than a million circulations), including 49 editions of Reader’s Digest, the world’s largest paid-circulation magazine, and sells approximately 30 million books, music, and video products across the world each year. It also owns the largest food web site in the world, Allrecipes.com, which has launched 15 web sites serving 20 countries. However, direct marketing was not enough to save a highly leveraged RDA in a post-Internet world where print media readership including that for magazines has been steadily declining. The company filed for bankruptcy during the global recession of 2009 but reemerged in February 2010. It is trying to reduce its debt substantially.42 Point to Ponder: How do you think TMBI should make the transition from direct marketing to CRM? What steps should the new management take?

Orlando Magic uses a CRM system to leapfrog NBA marketing. Source: Wikimedia by Jeff Kern.

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

128 RELATIONSHIP MARKETING RE-IMAGINED

Case-in-Point: Orlando Magic and the Statistical Analysis System SAS (Statistical Analysis System) is a software system developed by the SAS Institute for advanced analytics, multivariate analyses, business intelligence, and predictive analyses. The program provides solutions to the most critical challenges that marketers face today. It can mine, alter, manage, and retrieve data and conduct statistical analysis on it. Built on a comprehensive marketing technology platform, SAS helps firms enhance contextual customer communications based on insights derived from big data and analytics. Furthermore, it enables firms to evaluate and optimize the customer experience in the digital and physical worlds.43 SAS for sports helps teams find out how to vigorously engage with fans on league and team sites and social media channels. SAS for sports provides customer intelligence solutions that enable the teams to understand fans’ needs, preferences, and behaviors and retain and increase revenue from season-ticket holders and new fans. With SAS, the teams can also develop the fan experience and support high-value fans with more personalized and relevant communications, interactions, and promotions.44 Professional sports teams in smaller markets often struggle to build a big enough revenue base to compete against their larger market rivals. By using an SAS CRM system, the Orlando Magic has become one of the top revenue earners in the NBA. To increase revenue, what Magic needed was to better understand what fans wanted. The team decided to use SAS Analytics to facilitate predictive modeling, segmentation, and real-time decision making in order to make sure that fans receive the best value.45 The team has achieved great success by working on the resale ticket market to price tickets more effectively, forecast season ticket holders at risk of defection, and examine concession and product merchandise sales. These activities help the organization understand and offer what the fans want every time they enter the arena. Alex Martins, CEO of the Orlando Magic, claims that SAS helps them customize the fan’s experience in a robust way, which is one their biggest challenges. Since Martins’ leadership, the season-ticket sales have grown as large as 14,200, and the corporate sales department has achieved remarkable growth. Martins explains how SAS adds value to the Orlando Magic:46 This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.



Customer Relationship Management Analytics 129

We adopted an analytics approach years ago, and we’re seeing it transform our entire organization. … Analytics helps us understand customers better, helps in business planning (ticket p ­ ricing, etc.), and provides game-to-game and year-to-year data on demand by game and even by seat. … And analytics has helped transform the game. GMs and analytics teams look at every aspect of the game, including movements of players on the court, to transform data to predict defense against certain teams. We can now ask ourselves, ‘What are the most efficient lineups in a game? Which team can produce more points vs. another lineup? Which team is better defensively than another?’ … We used to produce a series of reports manually, but now we can do it with five clicks of a mouse (instead of five hours overnight in anticipation of tomorrow’s game). We can have dozens of reports available to staff in minutes. Analytics has made us smarter. Anthony Perez, Vice President of Business Strategy, highlights the importance of SAS in their achievements:47 SAS has helped us grow our business. It is probably one of the greatest investments that we’ve made as an organization over the last half-dozen years because we can point to top-line revenue growth that SAS has helped us create through the specific messaging that we’re able to direct to each one of our client groups.

Key Takeaways • Loyal customers are not necessarily profitable customers. There is a need to segment the customers according to their profit potential. • CRM is a comprehensive strategy and process of acquiring, retaining, and partnering with selective customers to create superior value for the company and the customer. • CRM has numerous benefits: it enables the marketer to measure and manage to optimize the cost to serve customers; profitability metrics can be tracked at the individual level; it empowers the firm to determine which customers are not This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.

130 RELATIONSHIP MARKETING RE-IMAGINED

•

•

•

•

•

profitable and “outsource” them, if necessary; it allows for fine-tuning of marketing campaigns at relatively low cost through experiments; it enables the measurement of ROMI; finally, CRM improves customer retention as well as customer acquisition and becomes more effective with more data over time. As much of our efforts are directed at generating differentiation through products or services that are not easily imitable, marketing inherently strives for imperfect competition. ­Relationship marketing in general and CRM via the use of CLV metrics in particular provide competitive advantage to firms that use them. The technical aspects of CRM implementation, although expensive, are relatively easy to handle. It is the human aspects of the CRM transformation that is challenging and time consuming. (a) The costs of serving long-life customers are not necessarily less, (b) they do not necessarily pay higher price premiums, and (c) they do not necessarily spend more. All of the above need to be qualified through the use of CRM. RFM, SOW, and PCV are traditionally used metrics to measure customer value. CLV has been claimed to be a superior metric because it considers the future value of a customer to the firm. CLV models range from basic to very complex. Strategies based on the use of CLV metric enable firms to measure the economic value of every customer and maximize their profits.

This document is authorized for use only in DR.ABHA WANKHEDE's PGDM-MKT/ MMS-MKT 21.01.2019 at Somaiya Vidyavihar from Jan 2019 to Jul 2019.