Celebros to offer product recommendation engine alongside site search
Site search started out with a basis of linguistic relevancy, but over the years it’s evolved into a merchandising tool that can make product recommendations based on business rules set by the merchant. As site search continues to evolve, site search technology vendor Celebros is keeping pace with that evolution by launching a new product that builds on its site search tool: a product recommendation engine driven by customer behavior, CEO Michael Crandall says.
The recommendation engine, now in beta tests with a number of retailers and being readied for launch in the next few months, makes product recommendations based in part on consumer behavioral analysis and shopping behavior analysis, as well as data from site search. “The product lives alongside search and takes advantage of it,” says Crandall, who also notes that it can be implemented on retail sites either with or without Qwiser, the Celebros site search tool.
Crandall says that an online retail operation as large as an Amazon.com gets so much consumer traffic that it likely has enough data to produce cross-sell recommendations that have a basis of statistical significance.
“But in the world of stores, not all of the SKUs have enough data on what was purchased to be statistically significant,” he says. But rather than populating cross-sell offers randomly where the data supporting the offer is not statistically significant, as some cross-sell tools do, the new Celebros recommendation engine looks to categories.
“Because our search is search and navigation, it has classified all of the SKUs in the shop,” he said. So if someone searches for a SKU without enough cross-sell data attached to it, the engine looks to the category into which it’s been classified to populate the cross-sell offer. “It makes recommendations based on more information than just a SKU-to-SKU correlation,” Crandall says.
In the beta tests, the recommendations populated by the engine and displayed in various parts of the test sites have produced double the conversions on cross-sell offers not selected by the engine, he says.
“Really good advertising is presenting to people what they are interested in,” he says. “What we think we are getting at is that not every shopper is going to look at the cross-sell box and be overjoyed, but a greater percentage of them will respond because what they see there is relevant and meaningful.”
Back...
