Online electronics retailer Outpost.com has yet to offer dynamic personalization on its popular Web site. Even so, data mining efforts provide key decision-making support for the company and its Web strategy, according to Edee Wollin, vice president of marketing.

For most online retailers, one of the biggest costs of doing business is marketing. In fact, Wollin says, whether marketing is working efficiently "is one of the single most important questions to answer regarding profitability." In December 1999, Outpost.com began using the data mining tools it acquired from SAS Institute, which employs neural network technology, to help the company assess the results of its advertising efforts. This is done both by examining those items that can be tracked directly to a direct response, such as an ad banner clickthrough, or by sophisticated neural net analysis of sales increases over expected sales that can be correlated to specific advertising events, Wollin explains. The solution was chosen because Outpost.com wanted something scalable enough to grow as rapidly as the retailer is growing.

Outpost.com uses data mining to create customer scores–rankings based on the customer’s buying habits–through a weekly regression analysis that reviews data collected during the previous seven days. The analysis provides "a sense of relative ranking–frequency of visits, recency and monetary metrics like lifetime spend," Wollin says. While personalization–using real-time analytics to present relevant offers while the customer is shopping–is expected next year, data mining helps the company identify and target its most frequent and most profitable customers. Outpost.com currently provides relevant offers based on products selected. "We want to interact with those customers in a way that may be perceived as more personal or more friendly–send them a note saying, ‘We know your TV is five years old. Would you like to take a look at the newest model?’ We’re building and reinforcing the customer relationship," Wollin says.