23% of e-retail transactions on Thanksgiving and Black Friday came from mobile devices, according to payments security firm ThreatMetrix. However, 15.5% of retailers say ...
Retailers are sharpening their focus on recommendations
Many e-merchants are pushing for product recommendations by 2011.
Chief Technology Editor
Web site personalization has come a long way in the past several years, bringing to fruition the early but unfulfilled promises of what was often called one-to-one marketing in the early days of retail e-commerce.
New personalization technology, now often referred to as collaborative filtering, has revived retailers’ interest and many are planning to deploy new online personalization systems by next year if they don’t already have one, says Susan Aldrich, senior vice president and senior consultant at research and advisory firm Patricia Seybold Group.
“The whole cross-sell, upsell thing is a key interest area,” Aldrich says. A survey her firm and the Institut Telecom Paris conducted earlier this year of 100 U.S. and European companies found that every respondent who didn’t already have a product recommendation engine in place was planning to deploy one by next year. Although the survey focused on businesses with an interest in product recommendations, Aldrich says she was still surprised to learn that every single respondent is moving to deploy a recommendation within the year.
The reason for the turnaround in the popularity of online personalized recommendations? It’s largely due to new technology that does away with the extensive labor required by earlier systems, which also were prone to slow down site performance, Aldrich says.
The old systems, she says, required online retailers to set long lists of business rules, which effectively made a site show particular products to site visitors who completed particular tasks on a site. “So retailers would have to write a lot of rules and keep them up to date. But if you had more than four rules, it could slow a site to a crawl—and yet you may have needed 50 rules.”
Collaborative filtering, she adds, takes a different approach. It uses modern web analytics to compile data about site visitors’ behavior that shows the steps, or site paths, they take that lead to an online purchase.
As visitors come to a site with a known shopping history, or land on a product page that indicates what they’re looking for, a recommendation engine may show them content that puts them on a shopping path that other shoppers with similar interests have taken before completing a purchase.
The recommendation engine is also usually designed to quickly modify its suggestions if a shopper doesn’t respond. “It’s like a global positioning system,” Aldrich says. “If a site visitor doesn’t take the recommended path, it will come back and recommend another path.”
In the study, Aldrich says, retailers indicated that their biggest expectation from web site personalization and product recommendation was that their efforts would generation additional revenue. “There is a very strong interest in deploying recommendations in the next two years in e-commerce, and you don’t want to fall behind—not when recommendations have such high impact on conversions and order size,” she says in the study.