Chapter 4

Identifying Customers

It wasn’t raining when Noah built the Ark.

—Howard Ruff

All enterprises use information about their customers to make smarter decisions. But for most traditional marketing decisions and actions, information is really needed only at the aggregate, or market, level. That is, any marketer needs to know the average demand for a particular product feature within a population of prospective customers, or the range of prices that this market population will find attractive. The enterprise then uses this information to plan its production and distribution as well as its marketing and sales activities.

But building relationships with customers necessarily involves making decisions and taking actions at the level of the individual customer, using customer-specific information in addition to information about the aggregate characteristics of the market population. This is because a “relationship” inherently implies some type of mutual interaction between two individual parties. We cannot have a “relationship” with a population or group but only with another individual. So the competitor trying to win with superior customer relationship strategies needs first to know the individual identities of the customers who make up the traditional marketer’s aggregate market population. Then the enterprise will make different marketing, sales, distribution, and production decisions, and take different actions, with respect to different customers, to create better experiences and increase customer value, even within the same market or niche population.

Individual Information Requires Customer Recognition

The essence of managing customer relationships is treating different customers differently; therefore, the first requirement for any enterprise to engage in this type of competition is simply to “know” one customer from another. However, identifying individual customers is not an easy process, and usually not a perfect one. Not many years ago, a British utility launched a December promotion to recognize its very best customers by mailing each of them a holiday greeting card. To the astonishment of its management, nearly 25 percent of these cards were returned to the company unopened in January. Apparently, many of the firm’s “most valuable customers” were actually lampposts. Until that time, this company’s management had equated electric meters with customers, comfortable in the knowledge that because they tracked meters, they also tracked customers. But lampposts don’t read mail or make decisions.

Most enterprises will find it difficult simply to compile a complete and accurate list of all the uniquely individual customers they serve, though some businesses and industries are more naturally able to identify their customers than others. Consider the differences among these businesses, and consider the advantage that would accrue to a company that’s able to identify individual customers and recognize each one at every contact:

• Telecommunications companies sell many of their services directly to end-user consumers. After all, to bill a customer for her calls in any given sales period, a phone company’s computers must track that customer’s calling activities—numbers connected to, time spent in each connection, day of week, and time of day. But even a cell phone company will likely make some sales to prepaid customers whose identities it can’t actually learn, because they buy their top-up cards in convenience stores or through distributors, and often prepaid customers want to maintain their anonymity. Such a firm may also serve a number of corporate clients whose end users are not specifically identified.
• Retail banks must know individual customer identities to keep track of each customer’s banking activities and balances. Historically, banks have been organized along lines of business, with credit cards, checking accounts, and home equity loans processed in completely different divisions. As a result, information about whether a branch banking customer is also a credit card customer often has not been readily available to either separate division. More and more banks are recognizing the need to coordinate and integrate information across product divisions, to produce a complete relationship profile of the customer accessible to all divisions in real time.1
• Consumer packaged goods companies sell their grocery and personal care products in supermarkets, drugstores, and other retail outlets. Although their true end customers are those who walk into the stores and buy these products, there is no technically simple way for the packaged goods companies to find out who these retail consumers are or to link their individual identities with their buying histories, except in some cases by using a “loyalty card” or other information-collection program. Recently, Infosys released Shopping 360, a sensor-based way for retailers and consumer packaged goods companies all to monitor and share information about what is bought off store display shelves by whom, if customers have given their individual permission.2
• Insurance companies can nearly always tell you how many policies they have written, but many cannot tell you how many customers they have or even how many households or businesses they serve. This is changing, of course, as more and more insurance companies recognize the need to base the organization and the reward structure for policy sales on customers.3
• A computer equipment company selling systems to other companies in a business-to-business environment may be able to identify the businesses it is selling to, but it is much more difficult for the firm to identify the individual players who actually participate in each organization’s decision to buy. Yet within any business customer it is these players—decision makers, influencers, specifiers, approvers, contract authorities, purchasing agents, reviewers, end users—with whom the selling company should be developing relationships. Thus some Web-based selling and contact-management tools are now able to help keep this information in a way that’s useful to the selling company.
• Carmakers, as well as state and local governments, have for decades recorded the current owner of each registered automobile by the vehicle identification number (VIN), visible through the front window of any car. However, even though the owner of each car can be determined, the cars belonging to each owner cannot. More recently, carmakers have begun relying on customer identification numbers, in order to tell whether two VINs are concurrently or sequentially owned by the same customer.4

Identifying customers, therefore, is not usually very easy, and the degree of difficulty any company faces in identifying its own customers is largely a function of its business model and its channel structure. In her book Customers.com, business consultant Patricia Seybold speculates that Microsoft probably has more than 100 million customers who have purchased its products through indirect channels (i.e., buying the software bundled on a computer, ordering it through a catalog, or picking it up at a store). And yet, for many years, the only end customers Microsoft knew about were the 90 enterprise accounts it maintained and the 25 million people who had sent in warranty registration cards. Then, in the late 1990s, as customers flocked to Microsoft’s Web site and interacted with the company, Microsoft began identifying greater numbers of its customers and collecting specific information about them individually, including e-mail addresses and product preferences.5

To engage any of its customers in relationships, an enterprise needs to know these customers’ identities. Thus, it must first understand the limitations, make choices, and set priorities with respect to its need to identify individual customers. How many end-customer identities are actually known to the enterprise today? How accurate are these identities? How much duplication and overlap is there in the data? What proportion of all customer identities is known? Are there ways the enterprise could uncover a larger number of customer identities? If so, which customer identities does the enterprise want to access first?

Step 1: How Much Customer Identification Does a Company Already Have?

To assess more accurately how much customer-identifying information it already has, an enterprise should:

• Take an inventory of all of the customer data already available in any kind of electronic format. Customer identification information might be stored in several electronic places, such as on the Web server or in the contact center database.
• Find customer-identifying information that is “on file” but not electronically compiled. Data about customers that has been written down but not electronically recorded should be transferred to a computer database, if it is valuable, so that it will be accessible internally and protected from loss or unnecessary duplication.

Only after it assesses its current inventory of customer-identifying information should a company launch its own programs for gathering more. Programs designed to collect customer-identifying information might include, for instance, the purchase of the data, if it is available, from various third-party database companies; the scheduling of an event to be attended by customers; or a contest, a frequency marketing program, or some other promotion that encourages customers to “raise their hands.”

Step 2: Get Customers to Identify Themselves

Sales contests and sponsored events are often designed for the specific purpose of gathering potential and established customer names and addresses. But to engage a customer in a genuine relationship, a company must also be able to link the customer to her own specific purchase and service transaction behavior. Analyzing past behavior is probably the single most useful method for modeling a customer’s future value, as we’ll see in Chapter 5, on customer differentiation, and Chapter 12, on analytics. So although a one-time contest or promotion might help a company identify customers it did not previously “know,” linking the customer’s identity to her actual transactions is also important.

Frequency marketing programs suit both purposes, providing not only a mechanism to identify customers, but also a means to link customers, over time, with the specific transactions they undertake. Such programs have been used for years to strengthen relationships with individual customers, but it’s important to recognize that a frequency marketing program is a tactic, not a strategy. It is an important enabling step for a broader relationship strategy, because a frequency marketing program provides a company with a mechanism for identifying and tracking customers individually; but this will lead to a genuine relationship-management strategy only when the company actually uses the information it gets in this way to design different treatments for different customers.

What Does Identify Mean?

Given that the purpose of identifying individual customers is to facilitate the development of relationships with them individually, we are using the word identify in its broadest possible form. What we are really saying is that an enterprise must undertake all of these identification activities:

Identification Activities

• Define. Decide what information will comprise the actual customer’s identity: Is it name and address? Home phone number? Account number? Householding information?
• Collect. Arrange to collect these customer identities. Collection mechanisms could include frequent shopper bar codes; credit card data; paper applications; Web-based interactions via Web site, blog comments, Facebook, or Twitter;6 radio frequency identification (RFID) microchips (such as E-Z Pass and ExxonMobil’s Speedpass); or any number of other vehicles.
• Link. Once a customer’s identity is fixed, it must be linked to all transactions and interactions with that customer, at all points of contact, and within all the enterprise’s different operating units and divisions. It is one thing, for instance, to identify the consumer who goes into a grocery store, but a frequent shopper program is usually the primary mechanism to link that shopper’s activities together, so that the enterprise knows it is the same shopper, every time he comes into the store.7 Also, if a customer shops online for a product but then contacts the company’s call center to order it, the relationship-oriented enterprise wants to be able to link that customer’s online interactions with her call-in order.
• Integrate. The customer’s identity must not only be linked to all interactions and transactions; it must also be integrated into the information systems the enterprise uses to run its business. Frequent flier identities need to be integrated into the flight reservations data system. Household banking identities need to be integrated into the small business records maintained by the bank.
• Recognize. The customer who returns to a different part of the organization needs to be recognized as the same customer, not a different one. In other words, the customer who visits the Web site today, goes in to the store or the bank branch tomorrow, and calls the toll-free number next week needs to be recognized as the same customer, not three separate events or visitors.
• Store. Identifying information about individual customers must be linked, stored, and maintained in one or several electronic databases.
• Update. All customer data, including customer identifying data, is subject to change and must be regularly verified, updated, improved, or revised.
• Analyze. Customer identities must serve as the key inputs for analyzing individual customer differences (see Chapter 12).
• Make available. The data on customer identities maintained in an enterprise’s databases must be made available to the people and functions within the enterprise that need access to it. Especially in a service organization, making individual customer-identifying information available to front-line service personnel is important. Computers help enterprises codify, aggregate, filter, and sort customer information for their own and their customers’ benefit. Storing customer identification information in an accessible format is critical to the success of a customer-centered enterprise.
• Secure. Because individual customer identities are both competitively sensitive and threatening to individual customer privacy, it is critical to secure this information to prevent its unauthorized use.

Technology is enabling enterprises to identify customers in ways never before imagined. Some enterprises still might use the old Rolodex card file system, but computer databases and sophisticated customer information databases are quickly supplanting these handwritten cards for the same reason that public libraries have abandoned their card catalog systems: because card catalog systems cost much more than their electronic counterparts and are available for search only in the physical library building. Sophisticated electronic data systems allow library patrons to search a library’s holdings from anywhere and help the library cut its own costs at the same time.

Integrated computer databases don’t just reduce costs. More important, they also help identify patterns that aren’t possible when the data is kept in filing systems or in separate data silos. The more the company integrates data from all corners of the enterprise, even including the extended enterprise, the richer in value the customer information becomes in planning and executing customer-focused strategies.

The end customer of an enterprise is the one who consumes the product or service it provides. That said, sometimes it is more of an indirect relationship, which makes it more difficult to tag the customer and link information to her. Sometimes, a product or service might be purchased by one customer and used by another member of the household or by the recipient of a gift. And as we discuss later, sometimes an end user will be an employee of a company while it is the company’s purchasing department that actually buys the product. Regardless of these intermediary relationships, however, it is the end user who is at the top of the food chain and the end user whose relationship with the enterprise is most important, because this is the person whose needs will or won’t be met by the product.

Customer Identification in a B2B Setting

A business-to-business (B2B) enterprise still must identify customers, and many of the issues are the same; but there are some important differences that merit additional consideration. For instance, when selling to business customers, the B2B enterprise must consider who will be on the other side of the relationship. Will it be the purchasing manager or the executive who signs the purchase order? Will it be the financial vice president who approved the contract? Or will it be the production supervisor or line engineer who actually uses the product? The correct way for an enterprise to approach a B2B scenario is to think of each of these individuals as a part of the customer base. Each is important in his or her own way, and each one should be identified and tracked. The greatest challenge for many businesses that sell to other businesses is identifying the product’s end users. Discovering who, within the corporate customer’s organization, puts a product to work (i.e., who depends on the product to do her job) is often quite difficult. Some methods for identifying end users include:8

• If the product consumes any replenishable supplies (e.g., inks, drill bits, recording paper, chemicals), providing a convenient method for reordering these supplies is an obvious service for end users.
• If the product is complicated to use, requiring a detailed instruction manual or perhaps different sets of application notes or even training, one way to secure end-user identities is to offer such instructions in a simplified, individually tailored format.
• If the product needs periodic maintenance or calibration or regular service for any reason, the enterprise can use these occasions to identify end users.

B2B firms use many strategies to get to know the various role players within the corporations they are selling to, from end users to chief financial officers—setting up personal meetings, participating at trade shows, swapping business cards, sponsoring seminars and other events, inviting people to work-related entertainment occasions, and so forth. But the single most important method for identifying the “relationships within relationships” at an enterprise customer is to provide a service or a benefit for the customer that can really be fully realized only when the players themselves reveal their identities and participate actively in the relationship. Thus, even though relationship marketing has always been a standard tool in the B2B space, today’s new technologies are making it possible more than ever before to manage the actual mechanics of these individual relationships from the enterprise level. In so doing, the enterprise ensures that the relationship itself adheres to the enterprise, not just to the sales representative or other employee conducting the activity.

Customer Identification in a B2C Setting

Can we identify—and recognize again and again—millions of customers? In the business-to-consumer (B2C) space, the technology-driven customer relationship management (CRM) movement has only recently made it possible even to conceive of the possibility of managing individual consumer relationships. But while managing relationships within the B2C space might be a relatively new idea, mass marketers have always understood that customer information is critical and that the possible ways of identifying customers are nearly limitless.

New technologies have made it possible to identify customers without their active involvement. ExxonMobil, the gasoline retailer, dispenses RFID microchips that can be carried around on the keychain of a customer who participates in its Speedpass campaign. When the customer drives up to a gas pump, the microchip device automatically identifies the customer and charges the customer’s credit card for the transaction. The customer is rewarded with a speedier exit from the gas pump (although she still must pump her own gas). The company, in turn, can identify each customer every time she buys gas at any ExxonMobil station and link that identification with every transaction.

Of course, few would deny that the Internet gave the biggest push to the customer relationship movement in the B2C arena. Not only did the World Wide Web provide tools to existing firms with which they could interact more effectively with their customers and identify an increasing number of them individually, but it also led to the creation of many new, Internet-based businesses with extremely streamlined business models based on direct, one-to-one relationships with individual customers, online.

As writer Stewart Alsop described the way Amazon.com led the way at the turn of the new century:

What Amazon.com has done is invent and implement a model for interacting with millions of customers, one at a time. Old-line companies can’t do that—I like Nordstrom, Eddie Bauer, Starbucks, and Shell, but they have to reach out to me with mass advertising and marketing. Amazon’s technology gives me exactly what I want, in an extraordinarily responsive way. The underlying technology, in fact, is revolutionizing the way companies do business on the Web.9

Customer Data Revolution

Clearly, in the Information Age, an enterprise can reach and communicate with individual customers one at a time, it can observe as customers talk to each other about the company, and it can follow strategies for its customer interactions that are based on relevant, customer-specific information stored in a customer database. The computer can now store millions of customer records—not just names and addresses, but age, gender, marital status and family configuration, buying habits, history, and demographic and psychographic profiles. Individuals can be selected from this database by one, two, three, or more of their identifying characteristics. CRM expert Stan Rapp has said that the computer has brought about “three awesome powers”: the power to record, the power to find, and the power to compare.10

• The computer’s power to record. In precomputer days, there would have been no point in recording by typewriter dozens of bits of information about each customer or prospect on thousands of index cards. Without the computer, there would have been no practical way to make use of such information. As computer data storage rapidly became more economical, however, it became possible and desirable to build up and use a prospect or customer record with great detail.
• The computer’s power to find. Selections can be made from the prospect or customer file by any field definitions or combination of field definitions.
• The computer’s power to compare. Information on customers with one set of characteristics can be compared to customer information using a different set of characteristics. For instance, the computer can compare a list of older people and a list of golfers.

For all its power, however, the truth is that when it comes to customer-oriented activities, the computer is an underutilized technology at most businesses—not because companies don’t want to use it, but because most customer data are simply not fit for use in an analytical database. The development of a database of customer information requires a data model—the tool required to bring data complexities under control. The data model defines the structure of the database and lays out a map for how information about customers will be organized and deployed.

At present, customer data are often duplicated in multiple operational databases. Multiple instances of data usually create data quality issues (think of how many times you’ve gotten mail from your bank with your name misspelled or with the wrong address), and these issues need to be addressed head-on if a customer relationship program is to be successful. This raises issues of data ownership and accountability, and can become a politically charged issue in the organization. But those companies that try to implement CRM without addressing this critical issue usually fail. It has to start with a single, complete view of each customer.

What Data Do We Need When We Identify a Customer?

After it has mined its existing customer databases and developed a plan to gather new customer information, the enterprise then decides how to tag its customers’ individual identities. Names are not always a sufficient customer identifier. More than one customer might have the same name, or a customer might use several different varieties of the same name—middle initial, nickname, maiden name, and so forth. To use a customer database effectively, therefore, it is usually necessary to assign unique and reliable customer numbers or identifiers to each individual customer record. It could be the customer’s e-mail address, phone number, a “user name” selected by the customer, or an internally generated identifier.

In addition to transaction details, other types of data generated from internal operations can make significant contributions. Information relating to billing and account status, customer service interactions, back orders, product shipment, product returns, claims history, and internal operating costs all can significantly affect an enterprise’s understanding of its customers. Directly supplied data consists of data obtained directly from customers, prospects, or suspects. It is generally captured from lead-generation questionnaires, customer surveys, warranty registrations, customer service interactions, Web site responses, or other direct interactions with individuals.

Directly supplied data consists of three obvious types:

In categorizing data contained in a customer database, it’s important to recognize that some data—stable data, such as birth date or gender—will need to be gathered only once. Once verified for accuracy, these data can survive in a database over long periods and many programs. Updates of stable data should be undertaken to correct errors, but, except for errors, stable data won’t need much alteration. In contrast, there are other data—adaptive data, such as a person’s intended purchases or even her feelings about a particular political candidate— that will need constant updating. This is not a binary classification, of course. In reality, some data are relatively more stable or adaptive than other data.

Why Is Identification Important?

Ultimately, of course, the central purpose of collecting customer information is to enable the development of closer, more profitable relationships with individual customers. In many cases, these relationships will be facilitated by the availability to the enterprise of information that will make the customer’s next transaction simpler, faster, or cheaper. Remembering a customer’s logistical information, for instance, will make reordering easier for her, and therefore more likely. Remembering this type of information will also lead the customer to believe she is important to the company and that her patronage is valued.

In order to make any of this work, however, it is essential for the enterprise to establish a trusting relationship with the customer, so she feels free to share information. A vocal privacy-protection movement—perhaps more active in Europe than in North America—has been energized by the increasing role that individual information plays in ordinary commerce and the perceived threat to individual privacy that this poses. However, both practical experience and a number of academic studies have shown that the vast majority of consumers are not at all reluctant to share their individual information when there is a clear value proposition for doing so and when they trust the company. Therefore, if a company can demonstrate to the customer that individual information will be used to deliver tangible benefits (and provided the customer trusts the enterprise to hold the information reasonably confidential beyond that), then the customer is usually more than willing to allow the use of the information. Trusting relationships or not, protecting customer privacy and ensuring the safety and security of customer-specific information are critical issues in the implementation of customer strategies and will be discussed in greater detail in Chapter 9.

Integrating Data to Identify Customers

The process of identifying customers in order to engage them in relationships requires that customer-identifying information be integrated into many different aspects of an enterprise’s business activities. It used to be that customer data could be collected over a period of time, and the customer database would be updated with revised profile and analytic information in batches. On weekends, perhaps, or late at night, information collected since the last update would be used to update the customer database. Increasingly, however, companies rely on Web sites and call centers to interact with customers, and this places a much greater emphasis on ensuring real-time access to customer-identifying information.

Enterprises must be able to capture customer information and organize it, aggregate it, integrate it, and disseminate it to any individual or group, throughout the enterprise, in real time. Technology is enabling enterprises to accelerate the flow of customer information at the most strategically timed moment. Enterprises strive for zero latency—that is, no lag time required—for the flow of information from customer, to database, to decision maker (or to a rules-based decision-making “engine”). The computer-driven processes of data mining, collaborative filtering, and predictive modeling will increasingly alter the process of forecasting how consumers behave and what they want,12 and, as more and more real-time interactivity continues to permeate all aspects of our lives, we can expect customers to demand more and more real-time service, which means enterprises will need real-time access to customer data.

In any service context, it is critical that an enterprise’s customer-facing people have ready access to customer-identifying data as well as to the records attached to particular customer identities. Making valuable customer information available to front-line, customer-facing employees, whether they work on board a passenger airliner, behind the counter at a retail bank branch, or at the call center for an automobile manufacturer, is an increasingly important task at all B2C enterprises.13

Borders Group uses a customer database and its Borders Rewards Program to give managers a consolidated view of its customers, including those from Borders Books, Borders Music, Waldenbooks stores, and the Borders.com Web site. With these kinds of data, Borders can discern whether customers who purchased books in its retail stores are the same ones who bought online. Borders is using data-integration technology to build the system and software to analyze the data and manage customer relationship campaigns based on the findings.14

Many enterprises underestimate the cost and difficulty of creating an integrated view of customer-identifying information. According to John McKean, author of Information Masters, testing an enterprise’s competency for using customer data requires that every aspect of the enterprise’s information environment affecting the efficiency of information flow be taken into account. This includes what McKean considers important, customer-facing functional areas, including direct marketing, customer service, and sales.15 McKean divides enterprises into three distinct categories based on their customer information competency: mass-market, transitional, and information mastery.

Professor Rashi Glazer clarifies the implications of an enterprise-wide view of the customer—what several authorities have called “one view of the truth” and others have called the “360-degree view of the customer.”

Summary

The first task to accomplish in building relationships with a customer is to recognize each one at every point of contact, across all products purchased or locations contacted, through every communication channel, over time. Doing this requires knowing the identity of each customer at every contact point in the organization.

Food for Thought

Glossary

Attitudinal data Directly supplied data that reflect attitudes about products, such as satisfaction levels, perceived competitive positioning, desired features, and unmet needs as well as lifestyles, brand preferences, social and personal values, and opinions.
Behavioral data Directly supplied data that includes purchase and buying habits, clickstream data, interactions with the company, communication channels chosen, language used, product consumption, company share of wallet, and so on.
Demographic data Directly supplied data that include age, income, education level, marital status, household composition, gender, home ownership, and so on.
Zero latency No lag time required for the flow of information from customer, to database, to decision maker (or to a rules-based decision-making engine).

1. “Achieving Customer Centricity in Retail Banking” (2006), Tibco Software, Inc.; available at: www.tibco.com/multimedia/solution-brief-achieving-customer-centricity-in-retail-banking_tcm8-2434.pdf, accessed September 1, 2010.

2. Infosys press release, “Infosys Technologies Launches Breakthrough Services for Retailers and Consumer Packaged Goods Companies,” Bangalore, India (July 31, 2008); available at: www.infosys.com/newsroom/press-releases/Pages/launches-breakthrough-services-retailers.aspx, accessed September 1, 2010.

3. Nadine Gatzert, Ines Holzmuller, and Hato Schmeiser, “Creating Customer Value in Participating Life Insurance,” working papers on Risk Management and Insurance, no. 64, Institute of Insurance Economics, University of St. Gallen (January 2009).

4. Richard Barrington, “Hard Lessons from CRM Experience: Six Mistakes to Avoid,” VendorGuru white paper; available at www.vendorguru.com, p. 4, accessed February 2, 2010.

5. Patricia B. Seybold, Customers.com: How to Create a Profitable Business Strategy for the Internet and Beyond (New York: Random House, 1998).

6. Michael S. Kenny and Will Yen, “Social Media: Examining Social Media Strategy and Architecture” (2009), Slalom Consulting white paper; available at www.slalom.com, accessed February 2, 2010.

7. Other information collection tools, such as the wireless, sensor-based tracking systems mentioned earlier, could supplement or replace a frequent shopper program. Infosys press release, “Infosys Technologies Launches Breakthrough Services for Retailers and Consumer Packaged Goods Companies,” Bangalore, India, July 31, 2008.

8. Don Peppers, Martha Rogers, Ph.D., and Bob Dorf, The One to One Fieldbook: The Complete Toolkit for Implementing a 1to1 Marketing Program (New York: Doubleday, 1999).

9. Back in 2001, Fortune columnist Stewart Alsop rightly pegged Amazon.com not only as a technology company when most relegated it to the more mundane role of e-tailer but as one of the only companies that had “mastered the use of technology in serving individual customers.” Stewart Alsop, “I’m Betting on Amazon,” Fortune, April 30, 2001, 48.

10. Stan Rapp, The Great Marketing Turnaround (Upper Saddle River, NJ: Prentice Hall, 1990).

11. Andrew R. Thomas, ed., Direct Marketing in Action: Cutting Edge Strategies for Finding and Keeping the Best Customers (Westport, CT: Praeger, 2007); David Shepard, The New Direct Marketing (New York: McGraw-Hill Professional Book Group, 1999).

12. Dolores Romero Morales and Jingbo Wang, “Forecasting Cancellation Rates for Services Booking Revenue Management Using Data Mining,” European Journal of Operational Research 202, no. 2 (April 2010): 554–562; Heung-Nam Kim et al., “Collaborative Filtering Based on Collaborative Tagging for Enhancing the Quality of Customer Recommendation,” Electronic Commerce Research and Applications 9, no. 1 (January-February 2010): 73–83; Rodolfo Ledesma, “Predictive Modeling of Enrollment Yield for a Small Private College,” Atlantic Economic Journal 37, no. 3 (September 2009): 323.

13. Don Peppers, “Customer Service at a CLIP,” Inside 1to1, June 10, 1999; available at: www.1to1.com.

14. Borders Group press release, “Borders Group Presents Long-Term Strategic Plan to Focus on Core Domestic Superstore Business,” Ann Arbor, MI, March 22, 2007; available at PR Newswire, www.prnewswire.com/news-releases/borders-group-presents-long-term-strategic-plan-to-focus-on-core-domestic-superstore-business-52175002.html, accessed September 1, 2010; Rick Whiting, “Borders Wants to Read Its Customers Like a Book,” Information Week, August 21, 2000, p. 34.

15. John McKean, Information Masters (New York: John Wiley & Sons, 1999).