February 1st, 2007
Demo 07: Aggregate Knowledge revs up discovery engine
Aggregate Knowledge is making retailer Overstock.com very happy. The company's Web service observes the anonymous behavior of people coming to the Web site and surfaces content relevant to what is on the page. It's a kind of recommendation engine, but unlike Amazon doesn't use any explicit personal data. The people who visited a page about sofa beds, for example, tended to visit these other pages, surfacing relevant alternatives.
During the holiday shopping season, Overstock saw an increase of 20 percent in sales due to the Aggregate Knowledge service, according to company CEO Paul Martino. He told me that the Aggregate Knowledge block, on the right side of the Overstock.com page, gets in excess of 20 percent clickthrough, and his company share in the revenue of items purchased that are bubbled up via the collective wisdom technology. In addition, the technology took only about two weeks to deploy at Overstock.com, which has over 800,000 items.
Aggregate Knowledge is also working with the Washington Post, delivering relevant content in real time based on reader behavior–"People who read this also read…." Martino has a bigger vision, the Collective Discovery Network, which is designed to deliver any kind of content from across multiple sites to users. If someone is reading a news story about a baseball team, the discovery engine could surface related merchandise, such as tickets and team apparel, from several sites that are part of the collective. The Collective Discovery Network won't available until later this year, Martino said.
Given the results Aggregate Knowledge has delivered so far from its behaviorally motivated product discovery engine, the company is on to something big as long as the algorithms hold up.
Dan Farber, editor-in-chief of CNET News.com, has more than 20 years of experience as an editor and journalist covering technology. See his full profile and disclosure of his industry affiliations.
