STORES Magazine - February 2009 - (Page 33)

WORTH WATCHING / MERCHANDISING Trendspotting Recommendation technology helps retailers capitalize on sales patterns BY FIONA SOLTES ome trends simply can’t be missed. Think Beanie Babies, Cabbage Patch Kids and Tickle Me Elmo, and you know full well how obvious they can be. S what’s working — and what isn’t. The system is completely automated, but if there’s a certain item or line that the retailer wants to promote, that’s possible, too. And it all happens in real time – much in the same way that trends can appear and then dissolve. The technology helps customers feel that the retailer knows what they want, sometimes better than they know it themselves. Who, how, what, where Even with those shoppers considered “low-engagement” with a low probability of return, “a lot of information is still given to us,” Selinger says. “We might know, for example, that they used Google to find the site, we know what they searched for and, if they purchased something, we know where that item was shipped to. And that can help us latch onto new trends. “If I bought the book ,“The Kite Runner,” during the holiday season,” he says, and the retailer later recommends another book by the same author, “that might be interesting to me.” The company’s most notable retail client is Sears. It targets large enterprises with “a pretty broad selection of products,” Selinger says, but clients range from those with 1,000 SKUs to “a few hundred thousand.” Perhaps one of the most appealing aspects of the technology, from the retailer’s perspective, is the fact that it can be up and running in a matter of days or weeks, even with “massive organizations.” And in these days of faster searches, shipments and trends, that speed is increasingly important. “That’s been where the rubber meets the road,” Selinger says. “When we engage with a customer, we’re not just engaging as a vendor. We become a partner, and that makes all the differStORES ence.” Fiona Soltes, who splits her time between “retail therapy” and freelance writing, lives near Nashville, Tenn. STORES / FEBRUARY 2009 33 Hiding in the shadows thrown by these mega-fads, however, are thousands of other small trends, and retailers have long hoped to latch onto them before it’s too late. Enter richrelevance’s personal recommendation technology, complete with a proprietary feedback loop. David Selinger, founder and CEO of San Francisco-based richrelevance, says it’s all about using data to move sales forward; in essence, it’s not just recognizing the trends as they’re starting to occur, but also being able to take full advantage of them once they have. That begins with a process called “ensemble learning,” which takes the place of more traditional tactics like collaborative filtering. With ensemble learning, the e-commerce technology runs more than 15 different types of real-time recommendations, such as recently viewed items and top sellers. Based on the previous behavior of the shopper (or others like him), the technology chooses the best two to four recommendation types for each page. Someone who previously purchased an electronic item, for example, may receive two sets of recommendations: One under the heading, “people who viewed this item also purchased this item,” the other, “people who viewed this item ultimately bought.” Another shopper who’s been looking for something of a certain brand, or in a certain price range, could receive additional recommendaWWW.STORES.ORG tions based on those criteria, as well. “Now, the whole concept of retail is way too big for us to wrap around a single narrative,” Selinger says. “Instead it’s: ‘Let’s try all of the things that make sense, and see which one of them works for your customers.’” Selinger, whose resume includes stints as vice president of software development and data mining at Overstock.com and the head of research and development for the data mining and personalization team at Amazon.com, says the problem with most recommendation technologies is they’re openloop systems: Once a suggestion is made, there’s no way to tell if that suggestion impacted the shopper or led to a sale. The richrelevance technology constantly incorporates what it’s discovering into the process, identifying http://www.overstock.com http://www.overstock.com http://www.Amazon.com http://WWW.STORES.ORG

Table of Contents Feed for the Digital Edition of STORES Magazine - February 2009

STORES Magazine - February 2009 Contents Executive Editor's Page President's Page Are You a Pusher or a Puller? What Shoppers Think Online Retail Satisfaction 10 Things You May Have Missed Numbers Worth Counting Full Price/Markdown Retail People Cover Story - Something’s Got to Give First Look Concept2Watch Online Partners Merchandising Sustainability Inventory Systems Drug Store Systems Business Intelligence Inventory Managment Online Marketing Supply Chain - Better Data, Better Decisions Returns Management - Identifying Fraud Data Security - Securing Intimate Data Anti-Shoplifting - Mall of Shame? Risk Management - Securing Consumer Confidence Loeb Retail letter ARTS Update Point of View NRF News Retail Crossword Retail Industry Calendar End Cap

STORES Magazine - February 2009

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.