Behavioral Segmentation for e-Tail Personalization

A multichannel retailer desires to develop a personalization strategy for customers who agree to receive e-mail communications containing imbedded Internet purchase appeals. Using customers' transactions history with the firm, four behavioral segments emerged from factor and cluster analyses. The segments were also distinguishable on the basis of attitudes toward shopping, determined from an attitude survey. A CART analysis provided an algorithm for classifying customers into the behavioral segments, based simply on their purchase history information. The segments were validated in a replication, which involved a survey with a new sample. Finally, different appeals were used in personalized messages tailored to two of the segments in an I-media experiment. The appeals resulted in higher sales response for one of the segments. 3 Although hyped as the eCommerce " Holy Grail, " true personalization at the individual customer level remains far in the future for most retailers. Certainly a degree of 1-to-1 communications between retailers and customers exists today, but customers rather than retailers typically initiate the exchange, and the actual personalization involved is fairly minimal (Peppers, Rogers, and Dorf, 1999). So-called " personalized " retailer communications are by and large reactive, rather than proactive or promotional. Reaching each customer with individualized promotional messages is largely infeasible for most retailers, despite the new information economics driven by the digital revolution (Peppard, 2000). Indeed, some have questioned whether personalization is even appropriate (Nunes and Kambil, 2001). Those customers who agree to receive persuasive communications through I-media, however, represent a potential source of new revenues for retailers having the desire and technology to address them. What can a retailer do to achieve a degree of personalization of promotional messages, without communicating individually with all customers? The answer would seem to lie in behavioral segmentation; that is, grouping customers with similar purchase behavior into segments for the purpose of targeting them with persuasive messages that are personalized at the segment level. Retailers potentially have in their databases vast amounts of cumulative information on the transactions of many of their customers. Not only can those databases contain transaction monetary value of customer transactions), but they may also record what was purchased on each occasion, including such data as styles, colors, and other item-specific information, 4 whether the item was purchased on promotion, method by which the purchase was made (credit card, check, cash), channel used (store, catalog, internet), and so forth. These are the " behaviors …