Data Best Practices For Spend Analysis

A DataFlux White Paper Prepared by:

David Loshin

Data Best Practices for Spend Analysis

Leader in Data Quality and Data Integration

www.dataflux.com 877–846–FLUX

International +44 (0) 1753 272 020

Corporate sourcing and procurement organizations should always look for opportunities to introduce efficiencies, reduce costs, as well as negotiate desirable terms with vendors and suppliers. These opportunities are revealed in a number of different ways, such as demand aggregation, improved supplier performance assessment, assurance of regulatory compliance, determination of rebates and refunds, and identification of noncompliant spend. All of these business benefits can accrue as a result of a process for reviewing and analyzing spend data. However, few companies have the ability to gain a comprehensive perspective of the products and services purchased and their associated providers. This confounds the ability to identify opportunities for improvement, and wasteful and duplicate spending can continue unabated. The difficulty in gaining this enterprisewide perspective is complicated by a number of factors, such as: 

There are often multiple systems used during the procurement process. With data spread across different data silos, it is difficult to consolidate spend data to provide summarizations across providers, products or commodity types.



Different vendors and providers use variant product and service identifiers and descriptions. The inconsistent naming and identification introduces challenges when analyzing purchases by product or product category.



Often, transactional data associated with purchasing is missing important characteristics that are used to influence and inform both the strategic and the operational decision-making processes for procurement.

Even when improvements such as negotiated product prices have been identified, ensuring that the negotiated savings can be achieved in practice require additional visibility into purchasing, supplier fulfillment and delivery data. Yet, according to an Aberdeen Group study, the top challenges for spend analysis include “poor data quality,” “too many data sources,” and “lack of standardized processes.”1 In essence, the biggest challenges to procurement improvements have to do with information, and the benefits of spend analysis can only be achieved when the spend analysis tools have access to the right data.

1

1

“Spend Analysis: Working Too Hard for the Money,” August 2007, Aberdeen Group.

In this paper, we look at the business drivers and organizational objectives of a spend analysis program, and then consider establishing performance indicators and associated metrics for managing the efficiencies and realizing cost savings. The paper then reviews spend analysis techniques along with the data management procedures necessary to enable the process. Last, we consider some of the most important challenges and associated techniques for driving a successful spend analysis program.

Business Drivers and Organizational Objectives Strategic sourcing incorporates best practices for evaluating the purchasing patterns and activities within the organization with the intent of identifying opportunities for improving the procurement process. These practices focus on examining the products and services that are bought, how those products are classified, who the suppliers are, how much is being spent on different types of items, and the terms under which these items are priced, purchased, and delivered. This process addresses specific business drivers associated with managing the way the organization spends money and seeking ways to reduce costs, improve operational efficiency, and better engage with the providers. Some techniques used for providing value using spend analysis include2: 

Demand aggregation – This is a process of consolidating purchasing requirements from across the organization into groups of similar items, thereby opening the possibility for volume discounts, reduced delivery costs, better purchasing terms and more control over specifications. Demand aggregation is also good for the supplier, who is able to capture a greater amount of organizational spend, reduce the cost of doing business, and streamline manufacturing and delivery efficiency.



Supplier assessment – This process evaluates supplier performance in terms of objective criteria such as credit scores and market value as well as scoring suppliers in terms of responsiveness, observance of terms, delivery times, pricing, product defects and warranties.



Regulatory Compliance – This supports compliance monitoring, especially with respect to controlled trade, OFAC compliance, doing business with suppliers from embargoed political regions, and tariffs.



Identify non-compliant (or “maverick”) spend – This process seeks to identify where individuals in the organization are purchasing products or services outside of the official procedures. Non-compliant purchases, such as picking up ink cartridges at the local office supply store instead of

2

For a comprehensive discussion, see the Spend Analysis and Opportunity Assessment wiki,

http://www.esourcingwiki.com/index.php/Spend_Analysis_and_Opportunity_Assessment

2

through the proper organizational channels, can result in paying a (potentially significant) premium for the purchased item. 

Fraud detection – This is intended to identify patterns of fraudulent spend, such as multiple invoices or payments to ghost companies.



Contracting strategies – These tactics help evaluate potential vendors, the types of items being purchased and their classifications. They also help in determining an appropriate procurement approach as well as positioning the purchaser for negotiation of favorable pricing and service.



Commodity analysis – This process looks at aggregated spending decisions to determine if the organization is buying commodity products in large quantities without even realizing it.

Defining Metrics As with any program intended to improve performance, it is necessary to examine the impacted processes from a business-driven perspective. Yet, according to a research report, a significant amount of negotiated savings can still remain unrealized.3 For example, even when the procurement organization has employed spend analysis techniques to negotiate better rates with suppliers, there is still a need to communicate those rates with everyone across the company. This helps make all individuals aware that their purchases should conform to a well-defined process in order to realize those negotiated savings. Therefore, it is critical to define the measures that will be used to indicate when the expected benefits are achieved. Let’s consider some key business drivers and map them to defined objectives as a way to understand the key characteristics that are indicative of better spend decisions, and examine metrics that will be used for those decisionmaking processes. For example, if the driver is cost reduction or cost savings, an objective metric might state “reduce materials cost by 10% within the next 18 months.” Other performance indicators that reflect improvement may include examples such as: 

Negotiated price reductions – As a result of demand aggregation, selected vendors can be approached to provide lowered prices in return for recognition as “best supplier” with longer-term contracts. This can be measured directly as the difference in negotiated price per product.

3

“Spend Compliance Management: Implementing and Sustaining Supply Savings,” December 2004,

Aberdeen Group

3



Communication of negotiated terms – To make all staff members aware of negotiated rates as well as defined procurement process, there must be a communication infrastructure in place for corporate awareness. Therefore, staff awareness goals can be set and measured.



Cost reduction – If the costs associated with procurement should be reduced through the identification of potential operational efficiencies, then specific goals for cost reduction can be set as program objectives. One example may measure time reduction for procurement of unique products.



Internal spend compliance – Once the negotiated rates and terms are communicated internally, it is important to make sure that staff members make their purchases through the approved processes and that they are purchasing the appropriate products through the selected vendors at the negotiated rates.



Supplier compliance – Once approved rates are negotiated, supplier performance can be monitored to ensure compliance with targeted pricing and delivery goals.



Demand reduction – Through improving product reuse and quality improvements applied to the purchased products, product lifetimes can be extended, thereby reducing the overall demand, and this can be measured as the reduction in requests for purchasing selected products.

What Does Spend Analysis Entail? Spend analysis encompasses the process of aggregating spend data together into a single framework in order to understand who in the organization is buying, what they are buying, from which suppliers, where the purchases are performed, and the different characteristics regarding the terms of the purchases (price, delivery, payment options, etc.). Even before you can analyze the many transactions, you face a more insidious set of challenges: 

Identifying the systems containing spend data



Collecting the spend data from the identified sources



Transforming that data into a usable format



Consolidating the data into a single repository

A first step is to ask a more basic question: Where does the data come from – and how can we make it more useful? There are many systems and subsystems, such as accounts payable, invoicing systems, purchase order management, supplier master data, expense report data, agreements and contracts, policies, purchasing card data, and others. There are many different underlying data sources, representations, and definitions, all of which must be identified, extracted, normalized, cleansed, transformed, classified and consolidated into a common data store. And since the most significant challenges

4

involve the collection, organization, and presentation of data for analysis, one might say that spend analysis is ultimately a data-driven process.

The Data-Driven Spend Analysis Process The spend analysis process is activated by the identification, normalization and consolidation of data, and consists of a number of distinct phases: 1.

Data collection – The first step is to identify the sources of data containing spend information. This may be no easy task. An organization may have multiple accounting systems, and there may be multiple processes and financial databases containing data associated with purchasing. For example, there may be documentation of product purchases through a purchase requisition and purchase order application, while vendor data is managed in a different set of systems. A survey of applications determines which underlying data sets contain data relating to procurement and spend. Once the data sources are identified, their structures and contents are evaluated, and the relevant records and attributes are extracted.

2.

Data normalization – Once the data sets have been extracted, data profiling can identify the similarities and differences to determine data mappings as well as gaps in completeness. A common model for representing spend data follows, and the appropriate transformations and normalizations are applied to convert the many data sets into a standard format.

3.

Data cleansing and enhancement – There is bound to be variance between the contents of the different source data sets – multiple versions of the same names for vendors, products and additional codes. Not only that, mergers and acquisitions may have modified corporate structures, introducing new organizational dependencies in the master vendor database. At this stage, data cleansing techniques are applied to parse and standardize vendor and product names and descriptions. Similarly, duplicate data can be identified and, in certain cases, eliminated. In addition, third-party data sources can be used for establishing the correct vendor and supplier corporate hierarchies. Other attributes associated with each vendor, such as “small business,” “minority-owned,” or “womenowned,” can be added also, which is important for analyzing certain aspects of regulatory compliance.

4.

Commodity mapping – Product data is particularly challenging in terms of cleansing and consolidation. Product data has wide variability and unpredictability when viewed across different business contexts, and is not necessarily suited to formatted pattern matching. There are differing standards for classification, presentation, and description. Product descriptions often contain various abbreviations and shorthand, yet these fields carry descriptive attribution embedded within free-formed text. Once

5

product data has been normalized, additional enhancements are applied to assign commodity codes to each product. Again, these enhancements can be facilitated using third-party data, such as the United Nations Standard Products and Services Code (UNSPSC). 5.

Classification and categorization – The actual analysis looks at transactions and spending patterns along a number of dimensions, either by corporate group, geographic region, supplier, and product type, to name a few. This analysis depends on the proper organization of the data, and at this stage, spend transactions that documented in the different data sources and have been consolidated are classified and organized along different dimensions of categorization. For example, “small paper clips” and “large paper clips” are both classified as “paper clips,” which is then categorized within the “paper fastener” category, and so on.

Figure 1: Activating spend analysis via data management techniques.  At this point the data is ready to be presented for the analysis and decision-making process. The different purchases can be aggregated to identify the most frequently used suppliers, which ones provide the best pricing, and how many accounts are active within the organization. Variance in commodity product pricing can be identified, while different purchasing characteristics can help in suggesting creative approaches for sourcing, ranging from reverse auctions for commodity pricing to the larger effort associated of a request for proposal (RFP) for more complex acquisitions. But long before any of this can take place, the data management challenges need to be addressed.

6

Meeting the Technical Challenges If it is clear that spend analysis will lead to improved operational processes, then, as with any analytical application, there is a dependence on well-defined data management practices. And if the top challenges involve the proliferation of data sources and the quality of the data, then incorporating these types of technologies with data management best practices will better enable a successful spend analysis program: 

Data integration/ETL – It would be reasonable to assume that spend data would be collected from the numerous data sources and centralized in a common data warehouse. Therefore, the spend analysis solution should include data integration tools and utilities to simplify the extraction of source data, transformation into a common format for analysis, and loading into the data warehouse.



Data quality – Spend analysis hinges on standardized and cleansed vendor and product data; traditional data quality management tools can help with parsing, standardization, and normalization of data, as well as exact and approximate matching algorithms that help in duplicate identification and elimination. In addition, inspection, monitoring, and reporting of compliance with data quality rules will better ensure high quality spend data in preparation for analysis.



Master data management (MDM) – Any consideration of unique identification of either products or suppliers suggests managing a master directory or repository of this data to support the analysis process. Tools supporting analytical MDM may already combine data integration and data cleansing with defined data models representing products and parties.



Data enhancement services – Third-party data vendors provide value-added enhancement services, either by supplying data or providing services for appending additional data attributes and characteristics.



Process standardization – Of course, the absence of well-defined processes for managing the data management practices diminishes the value of any acquired technology.

No matter how advanced the technology is for analyzing spend data, the identified opportunities for improvement will only be accomplished when the management supports the organizational change management. Defined performance objectives, specific criteria for success, accompanied by metrics that monitor how well the organization adapts to those improvements will determine whether the expected savings can actually be realized.

7

Conclusion Corporate sourcing and procurement organizations seeking process improvement can implement a spend analysis program to identify opportunities to reduce costs while improving counterparty relationships and establish desirable prices and business terms with suppliers. Spend analysis not only is used to find these opportunities, but the additional visibility into purchasing, supplier fulfillment, and delivery data also helps ensure that the negotiated savings can be achieved in practice. Many opportunities lurk hidden in spend data: the ability to aggregate purchases, identify preferred suppliers, negotiate better rates and prices, improve terms of engagement, support regulatory compliance, identify fraud, and even drive different procurement strategies. But spend analysis is particularly dependent on high-quality data collected from across the organization. There are challenges to implementing an effective spend analysis program, such as the proliferation of systems containing spend and procurement data, the variance in product names, product descriptions, and vendor names, and the need to enhance the data with additional characteristics often missing in the source. Spend analysis is essentially a data-driven process, from the identification of the appropriate data sources, data extraction, transformation, cleansing, and normalization, along with the potential need for master product and master vendor directories. At the same time, implementing best practices in data management and in performance management will most effectively support the analysis and improvement cycle. To benefit from spend analysis, a good approach is standardizing and managing the data management and analysis processes, so that individuals can exploit actionable knowledge while continuously measuring the success of the program.

8