The call for an audit of Facebook’s metrics comes a week after the social network acknowledged inflating its video metrics.
For personalization to work, a site’s merchandise must be categorized carefully and the site needs a lot of repeat traffic, a J. Crew’s executive says.
Dave Towers, director of customer experience for e-commerce at J. Crew Inc., readily admits his company is far from being the leader in personalization. He-and most others-grant that distinction to Amazon.com. Yet Towers did his best to convince his audience at the eTail Conference in New York this week that his chain is thoughtful in applying such technology. “To us, personalization means providing targeted content to customers based on their profiles in order to improve the experience and reduce clicks on the site,” explained Towers. “We use it as a tool to improve ROI.”
It seems the tool is working for J. Crew. Last year, its web site generated $107 million, compared to $500 million for its stores and $177 million for its catalog. Furthermore, he claimed, the site is profitable, sports an impressive conversion rate of better than 5%, boasts an average order of $125, and registers 30 millions sessions annually.
But the productive and profitable use of personalization, said Towers, will vary by type of retail web site. The greater the number of products available on the site, the greater the need for effective personalization to reduce clicks necessary to find desired products. However, a wealth of products that are not clearly categorized in distinctive groups will render personalization less useful, because personalization’s strength is in recommending to customers products that fit into the same category from which they made an earlier on-line purchase. And personalization is best suited for web sites that have a lot of repeat traffic.
Implementation techniques for personalization also vary widely from site to site. Some sites, such as Amazon, favor “explicit” personalization programs that openly recommend merchandise closely related to something the web shopper has recently purchased. The technique is likely to result in a more actionable message, but it also opens the risk of offending the shopper if the recommendation is off base. Other sites favor implicit techniques that merely show the related merchandise without explaining why or making a recommendation. It has the very opposite pro-con equation.
Personalization, said Towers, also varies by when it is implemented. In a static approach, noted Towers, personalization is used as the consumer enters the site; it does not respond to how the shopper is navigating the site. By contrast, real-time personalization feeds content to customers that attempts to respond to what those customers are doing on the site at that time. Even the method of data collection for profiling varies, said Towers, from that provided by the customer responding to an on-line survey to that assembled by analyzing a customer’s pattern of web usage.
All of these parameters must be examined before a retailer can determine whether and what type of personalization program meets its needs. But, cautioned Towers, personalization technology is certain to undergo dramatic change in the next five years. Specifically, Towers referred to the development of XML technology which he believes will eventually replace HTML formats. XML allows web sites to access and deliver much more data than can presently be presented via the web. In addition, he foresees a major change in personalization once intelligent and automated browsers are marketed in the next several years. Using that technology, consumers will enter detailed profiling data into their browsers’ programs, which in turn will search through massive databases on web sites and automatically retrieve products and content that matches the computer user’s profile. “Don’t get too comfortable with whatever you are doing with personalization now,” Towers cautioned, “because the technology is about to change.”