In these examples, we see one case of long-term loyalty and affinity, another case of loyalty because of perceived customer advantages, and a third case of zero loyalty. The first two cases are acceptable for retailers, because they result in long-term customers; in the case of the latter, it creates challenges for the retailer. (Add in a huge lottery jackpot, and on Wednesdays you may find us steering clear of one particular chain of gas stations!)

One important component of a retailer's CRM program is a solid customer database (also known as a data warehouse). In this "warehouse" are stored pertinent customer information, as well as a complete record of all transactions. This includes how responsive customers are to special promotions, sales, etc., whether they use a frequent customer card, how much they spend, what types of items they spend, and the apparent preferences of the customer. Furthermore, any additional customer demographic information that can be gleaned can be of great help in future individualized marketing campaigns (which may include emails, targeted mailings, etc.).

E-commerce firms practice database management as well. In fact, they have an advantage because all of their sales are already recorded in a database, along with customer information.

Gathering customer information is not a new phenomenon. As far back as I can remember, Radio Shack asked for a complete address when buying even the cheapest little cable or adaptor. Service Merchandise once asked for residential phone number so that it could then access its criss-cross phone directory, and thus keep a record of sales (and then send you a catalog). Even Home Depot has asked me for my ZIP code, which reveals little other than the general area where their customers reside.

Customer loyalty programs today are commonplace. They usually offer some "perk" or reward for continued patronage, but in the process allow the retailer to gather a long line of historical transaction data. The benefits are thus perceived to be positive for customers and retailers alike. A more recent twist is the customer cards that supermarkets offer. These have been used in a recent test to offer students a dynamic pricing model (e.g., special deals for an audience of one), and also provide in-store electronic reminders via a high-tech shopping cart.

Other stores have gathered enough information about users that they have been able to connect credit card and/or checking account numbers with a driver's license, which is tantamount to knowing virtually everything about the customer. This allows the firm to compile information without the customer actually having had to enroll in some sort of program.

The biggest concern about all this information gathering is customer privacy, for great potential exists for abuse if information were to fall into the wrong hands. Will the information be sold or shared with other firms? How will the information be used? Are cookies being placed on customer computers that send web history data?

The issue has been further complicated by what appears at the surface to be customer input as to whether information is collected. Called "opt-in" programs, they appear to be volitional by the customer. However, some programs shroud their intentions and bury them far inside a "terms of use" agreement that few people read or understand. Thus, many customers unwittingly opt-in to a program, not really knowing what is going to happen to their information.

Data Mining

All the databases in the world are useless unless the information is mined. Bill Gates, in his book "Business @ The Speed of Thought," calls this "pivoting" in the database. In other words, the analysts "pivot" on a certain key variable, and then crunch the data to discern patterns or golden nuggets of information. It's the difference between simply understanding that most of your customers are from Minnesota, and also knowing that those Minnesotans purchase a lot of add-on products (and which ones). The result allows retailers to make very specific, targeted campaigns with certain products to certain customers.

Among the many uses of data mining are:

• Market basket analysis, meaning the items in the transaction are analyzed to look for patterns. This can be helpful for store layout, and even multiple placings of products (e.g., batteries)
• Identifying customer segments you did not know exist, even by day-times.
• Identifying the best customers, and recognizing the so-called 80-20 rule.
• An RFM analysis will help the retailer understand how recent a customer puchased, how frequently they buy, and how much they buy. Again, this is helpful for identifying the heavy hitters.

The ultimate goals of a CRM program are to retain the best, most-profitable customers, convert other customers into high-profit customers, and get rid of unprofitable customers. While this sounds rather exclusionary, the retailer must make decisions about whether it wants to serve customers that are actually dragging down operations.

As for the first two groups of customers (the "keepers"), the key is building on the "audience of one" strategy mentioned above. Direct, 1-on-1 marketing can help convert average customers into high-profit customers, and can then keep them coming back. As for the unprofitable customers, management will often send subtle messages to them that they are not valuable, and thus implicitly discourage them from coming back. This, of course, can backfire if not handled properly, for it can make the retailer look elitist and smug.