See Description

Christopher Groomes

Professor Flores

American Military University

INFO531

August 31, 2014

Asking the Customer by Asking the Database CASE STUDY

What’s the best way to find out what your customers want? If you’re a large business with millions of customers, it’s impossible to ask each one face-to-face. But thanks to modern data management and data mining technology, you can “ask” each one by mining your customer database. The results are dazzling, and perhaps a bit unsettling as well.

Customer databases typically contain data such as a customer’s name, address, and history of purchases. These databases include records of the company’s communication with its own customers or customer “lists” purchased from other organizations, including charity donation forms, application forms for free products or contests, product warranty cards, subscription forms, and credit application forms. Today these databases are starting to include customer data gathered from social media, mobile, Web, and e-mail transactions.

Until fairly recently, large companies, such as Forbes, did not look closely at their own data. Forbes is an American publishing and media company that publishes the bi-weekly Forbes business magazine and maintains an extensive Web site. Its worldwide print and online readership numbers over 45 million people. Forbes’s business model relies heavily on advertising to supplement revenue from paid subscriptions, so it is constantly looking for ways to help its advertisers reach Forbes readers more effectively.

For many years, the various entities in the Forbes media empire employed third-party research services to analyze their customer data. These services compiled information about Forbes readers by selecting and analyzing subsets of the reader population as a way of learning about the entire readership. Decisions were made on the basis of what could be predicted about the “average” reader. But readers are individuals, and what Forbes and its advertisers really wanted was to find out what each individual customer was doing with Forbes publications.

Enter corporate databases and data mining. Forbes did maintain extensive data on its individual subscribers and Web site visitors from magazine subscriptions and visits to its Web sites. It just needed to make better use of the data. Management realized it could actually learn details about each of its individual readers by examining Forbes’s entire reader population, using the data it had already accumulated on a regular basis.

Forbes started to use SAP BusinessObjects software to analyze its own readership data, examining variables of greatest interest to its advertisers. Forbes claims it can now understand each individual who interacts with its brand. Whether that person is a subscriber or registered Web site visitor, Forbes has some knowledge of that individual’s demographics, values, and lifestyle as well as how that person has interacted with Forbes over the years. These details help Forbes’s advertisers target their campaigns more precisely and also help Forbes’s publications increase their circulation.

Monster.com, one of the world’s largest job listing sites, used business intelligence analytics to scale back its broad-based brand advertising in favor of a more targeted multichannel approach. Most of Monster’s revenue comes from employers who pay to post job listings and to search its resume database. (Job seekers can post resumes and search listings free of charge.)

A typical campaign starts with e-mail, but in the past Monster had been sending generic messages about itself and its services to large groups of companies. The recent recession and high unemployment rates inspired the company to look for a more cost-effective approach.

Monster now tries to find new employers and candidates using much more personalized e-mail, direct mail, social engagement, and prioritized telemarketing using IBM’s Unica enterprise marketing management tools. A typical campaign now starts with a personalized e-mail message to targeted segments, Monster’s best prospects being human resources decision makers in large companies in growth industries such as health care and technology. Monster uses SAS statistical modeling software to identify existing and prospective customers who are most likely to purchase job listings and other services in a specific quarter. A Unica marketing database maintains data on when the e-mail campaigns ran, e-mail recipients, who responded to the e-mail messages, and who clicked through. By analzying these data, Unica is able to generate mailing lists based on criteria such as past response behavior and an “opportunity score.”

Monster’s business intelligence (BI) analytics tools examine attributes such as industry, company size, and location to score the data and target a subset of around 1,000 human resources (HR) executives identified as top prospects who might merit special treatment, such as an expensive direct mail promotion or even a gift. For example, to promote its new Power Resume Search service, Monster sent the leading prospects global positioning system (GPS) devices, combined with promotional messages describing the service as a GPS for finding job candidates.

Monster also uses its Unica data to initiate interactions with select prospects through LinkedIn and other business-oriented social networks. If it has a target list of 1,000 executives, Monster tries to engage 50 to 100 of them using social media.

Behaviors tracked through Unica also help Monster’s sales force target prioritized telemarketing follow-up calls. For instance, any customer who has opened and clicked on more than one e-mail will most likely receive a call.

Diapers.com, the largest online specialty retailer for baby products, started out in 2005 as a bootstrap operation by entrepreneurs Marc Lore and Vinit Bharara. At that time, the company didn’t have any historical data to predict how new mothers would behave, and decisions were based on the co-founders’ firsthand knowledge that new parents would be attracted to overnight shipping of essential items and reliable customer service. The company’s business strategy focused on cultivating customer loyalty by touting cheaper items such as infant formula and baby powder to entice customers into purchasing higher-end goods such as car seats or baby swings that could be conveniently shipped in the same box.

Diapers.com’s parent company Quidsi (owned by Amazon.com) now uses predictive analytics with more than five years of historical data on customer spending to calculate how much each buyer will spend over that person’s lifetime as a customer. Geographic location and product choices are important variables. The analytic data drive the company’s marketing budget for different customer demographics.

Like other retailers with low profit margins from online sales, Diapers.com does not make anything off the first purchase. If a customer does not shop on a repeat basis, Diapers.com doesn’t believe that customer is worth having. According to Marc Lore, once Quidsi calculates the profit it will make from each customer over a lifetime, it knows how much it’s willing to spend to acquire and retain that customer. In 2010, the average Diapers.com customer cost about $40 to acquire, but that person would contribute an average of $70 to the company’s bottom line over the lifetime of purchases from the company.

Target, the second largest U.S. discount retail chain, has been able to take predictive analysis of customers to new heights by incorporating scientific findings about habit formation. For decades, Target has amassed vast quantities of customer data. It assigns each shopper a Guest ID number, which is a unique code for keeping track of everything an individual customer purchases. The Guest ID is linked to data about whether a shopper uses a coupon or credit card, mails in a refund, fills out a survey, responds to an e-mail, visits Target’s Web site, or calls Target’s customer help line.

Also linked to the Guest ID is demographic information, such as a customer age, marital status, number of children, residential location, whether the customer recently moved, what credit cards the customer uses, and what Web sites the customer visits. Target can purchase data about your job history, ethnicity, magazine subscriptions, whether you’ve declared bankruptcy, what brands of coffee and paper towels you prefer, your political leanings, charitable contributions, and the number of cars you own.

Target’s predictive analytics department is further pushing to increase sales by using findings about habit formation, which show that 45 percent of the choices people make every data are based on habits rather than conscious decision making. Purchases for mundane products such as soap, toothpaste, and paper towels are typically made habitually, with no complex decision making. For such habit-driven purchases, special displays, product promotions, and coupons have little impact. But when customers are going through a major life event such as moving to a new town or expecting a baby, their shopping habits become more flexible and open to intervention by marketers.

Target was able to mine its customer data to identify about 25 products, such as unscented lotion and cotton balls, that, when analyzed together, resulted in a pregnancy prediction score, with an estimated due date, for each shopper. Shoppers with high prediction scores were more open to purchasing a whole array of products from Target that they had previously bought out of habit from other retailers. A Target statistician created a pregnancy prediction score for every female shopper in its national database and was able to come up with a list of tens of thousands of women who were most likely to be pregnant. If these women could be enticed into a Target store to buy baby-related products, they might be open to buying groceries, toys, and clothing from Target as well.

When Target started sending coupons for baby items to customers according to their pregnancy prediction scores, the company quickly found out that this made people uncomfortable, even though it was strictly complying with federal and state privacy laws. So instead of sending people with high pregnancy prediction scores books of coupons solely for baby items, Target tried to make the baby ads look random by mixing them with ads for things it knew pregnant women would never buy. For example, Target might put a coupon for wine glasses next to infant clothes to make it appear that all the products were selected by chance. Target found a pregnant woman will use its coupons as long as she thinks she hasn’t been spied on.

Target may want to keep its pregnancy prediction score formula a closely guarded secret, but there are many companies selling pregnancy and baby-related products that might be interested in purchasing this methodology. Under current privacy laws, there’s nothing to prevent Target from sharing such information with retailers and other organizations that are not part of the Target family but want to provide customers with special offers of their own.

Sources: Charles Duhigg, “How Companies Learn Your Secrets,” The New York Times Magazine, February 16, 2012; Kashmir Hill, “Could Target Sell Its Pregnancy Prediction Score?” Forbes, February 16, 2012; Doug Henschen, “IT Meets Marketing,” Information Week, April 11, 2011; “Forbes: SAP BusinessObjects Software Provides New Marketing Insights,” www.sap.com, accessed April 21, 2011.

(Laudon 243-245)

Laudon, Kenneth C., Jane Laudon. Management Information Systems: Managing the Digital Firm, 13th Edition. Pearson Learning Solutions, 01/2013. VitalBook file.

Chapter 6 Case Study 1