Nexchange operates more than 200,000 e-commerce sites for a range of retail affiliates. But behind its online storefronts lurks an ugly secret: Over two-thirds of shoppers abandoned their carts in the middle of a visit.
The reason, CEO Del Ross discovered, was that customers weren't getting answers to questions such as where to find what they wanted or how to check out. Rather than rework his "help" pages, Ross decided to incorporate natural-language search technology from eGain in San Francisco. The idea is to make shopping easier for customers by enabling them to ask questions in plain English.
"Being able to talk is a very human thing," says Ross. "It makes the sterile online shopping environment a little more human; it increases the comfort level."
And what a difference a little comfort makes: Since Atlanta-based Nexchange implemented natural-language search technology, the company's abandoned shopping-cart problem has decreased by 25 percent to 30 percent, according to Ross.
Nexchange isn't the only company lending its ear to natural language. Dozens of e-commerce sites are eagerly implementing natural-language search engines in hopes of humanizing the shopping atmosphere and boosting sales.
But a natural-language search is not a one-step recipe for e-commerce success. Making the technology work as advertised requires well-planned implementation, lots of research and continuous tweaking. A sloppy implementation will not only make a site look dumb, it will also anger your customers.
Simply put, natural-language processing is technology that lets people interact with machines in conversational English. Originally developed in the late 1950s, the technology didn't become a household word until 1997, when Emeryville, Calif.-based Ask Jeeves unveiled an Internet search engine that used natural-language processing instead of the standard keyword-based approach. Whereas the keyword method retrieves every document containing the exact keyword searched for, a natural-language search "infers" the meaning of a query and finds only the most relevant items.
The market for online search is huge. According to Jupiter Communications, searching is the second-most-popular activity among Web users, following e-mail. Despite the popularity, though, keyword searches are far from perfect. In a review of Web sites, Forrester Research (FORR) discovered that a mind-boggling 92 percent of searches failed to find relevant information or to arrange the results in a meaningful order.
Consequently, "everyone's trying to get away from keyword" searches, says Paul Hagen, an analyst at Forrester Research. Web businesses in particular are anxious to improve their sitewide search capabilities, so consumers can more easily find the products they're looking for. Many are seizing on natural-language searches as the answer.
Big-name players like Dell (DELL), E-Trade, Microsoft (MSFT) and Nike have installed natural-language search elements on their sites. And companies like AnswerLogic, Delphes and KachinaNet.com have been quick to follow Ask Jeeves' lead and offer natural-language search products to such online businesses.
