HOT PROJECTS: Retail

"Our marketing department is extremely zealous. They like to slice it fine," says Van Rhodes, manager of marketing decision support at clothing cataloger Newport News. Through the 1990s the company worked to refine and supercharge its data warehouse, using SAS software to build a decision support front end for marketing personnel, instead of leaving them to create their own ad-hoc reports, a time-consuming process. For the final piece of the puzzle, Newport News worked with SAS partner Tricision to develop a more refined predictive model to steer dozens of different catalogs to thousands of different customer segments. The models dig deeper than the typical counts of recency, total business, and frequency. "Before we had the data warehouse, we had to analyze those [segments] in fairly broad buckets," Rhodes says. The new track-and-send model went in place in January, and the improvements in gross revenue and reduction in marketing support head count have Rhodes predicting a payback on investment within a year and a half. "Tricision was able to build models with significant lift of 10 percent per catalog, which is huge for us--we would have been happy with 1 percent." Michael Relichspent a great deal of company money on a data warehouse project, but store planners shunned the somewhat arcane analytical tools in their retail management package. Relich, CIO of retailer Wet Seal, boosted adoption of the analytical applications by implementing MicroStrategy's business intelligence tools and improving the accessibility of preprogrammed reports. So when a Wet Seal subsidiary president mandated its use for her entire business unit, demand improved considerably. "I wanted to increase usage, and I found out we could accomplish these things from the data warehouse" instead of buying more expensive new software, Relich says. By exposing preconfigured reports as desktop icons for planners and managers, the company has eliminated much of the expense associated with redundant programming work on ad-hoc reporting projects, and more important, has a better understanding of how product demand differs in each of its hundreds of stores. "I know we're making better allocations, because they're getting data they haven't gotten before," says Relich, who is confident that the improvements will help reverse a downward sales trend. Nails and hammers, saws and blades ...they're obvious pairs. It's the more obscure product relationships that can be the most lucrative, however, says Nathan Miller, e-commerce marketing manager at hardware retailer Northern Tool. "There were two ways to go about [finding those relationships]. We could either annually look at the top 3,000 best-selling products and assign cross-selling products based on our perceptions, one by one, or use Coremetrics to derive the data for us," Miller says. After a successful trial with the sales analysis tool during which customers who investigated recommended cross-sells doubled conversions and increased order size 5 percent over those who did not, the company was sold. It has committed to a full long-term rollout. Northern Tool searches for the top-seven products whose sales are associated with another item in its online catalog, then cycles through recommendations of the next three not already listed as accessories or spare parts for the product. Miller says the pace of improvement in online sales will lead to $3 million in additional revenue on an annualized basis.