Computerworld - BOSTON -- Clickstream data analysis showed Lands' End Inc. that "search" was the No. 1 destination for visitors to its online commerce site. But much to its frustration, further investigation revealed that shoppers came up with "no results found" more often than the company would have liked.
So Dodgeville, Wis.-based Lands' End, which was purchased in June by Sears, Roebuck and Co. (see story), joined the growing number of retailers that are abandoning their old homegrown or packaged search engines and spending money on new software that they hope will help to improve search engine performance and, ultimately, boost customer satisfaction and sales.
Several retailers attending last week's eTail 2002 conference here said they have implemented, are in the process of installing or hope to soon explore more accurate search engine technology that's being put out by vendors such as EasyAsk Inc., Endeca Technologies Inc. Mercado Software Inc. and Verity Inc.
"You cannot take your eye off the basics; you need to do them well," said Sam Taylor, a vice president at Lands' End.
Harley Manning, an analyst at Forrester Research Inc. in Cambridge, Mass., said increasing numbers of clients have shown interest this year in search engines, which start at $100,000 but typically cost $200,000 and up.
"It seems like search is so fundamental that people should have been focusing on it all along," Manning said. "The reality of the situation is that there was a great assumption that search was actually working just fine."
Many retailers don't even have search capabilities on their sites, said Geri Spieler, an analyst at Stamford, Conn.-based Gartner Inc. In a report published earlier this year, Spieler listed the lack of search capabilities as one of the most egregious mistakes many retailers have made when designing their online businesses.
Cathy Hotka, vice president of IT at the National Retail Federation in Washington, said retailers historically have faced difficulties when trying to come up with product definitions. "It's one thing if a shopper is looking for a commodity item like a 2-liter bottle of Pepsi," she said. "It's another thing if they're looking for a slinky black skirt."
That's the sort of problem that The Talbots Inc. in Hingham, Mass., sought to address with its enhanced search engine from EasyAsk. Talbots' old keyword-driven engine was limited to the descriptive terms employees entered into the system, said Jan Linert, manager of online marketing at Talbots. Its new EasyAsk engine can handle natural-language searches and can even process a request with multiple elements, such as "purple sandals under $50."
Browsers Become Buyers
Linert said Talbots has already seen results since launching the search engine in March. The average order size for shopping carts that hit the EasyAsk search engine vs. the old one has increased 18%, and the number of shoppers who search and then purchase items has risen 34%, he said.
Several retailers said they didn't realize the extent of their search engine woes until they implemented better data analysis tools. Lands' End even did usability studies at a market research facility, watching shoppers try to perform a list of specified tasks.
Michael Crotty, vice president of marketing at Neiman Marcus Online in Dallas, said his company knew anecdotally that its search engine wasn't ideal. But it wasn't until it used data analysis tools that it realized visitors were just leaving the site once their requests came up empty. Neiman Marcus researched eight search engine vendors before settling on one last week. Crotty declined to name the vendor.
David Kardesh, CIO at The Spiegel Group in Downers Grove, Ill., home to the consolidated IT department for retailers Spiegel Catalog Inc., Eddie Bauer Inc. and Newport News Inc., said his company gave each of its four contenders a copy of its database so they could test their search engines.
Spiegel settled on the search technology from Cambridge, Mass.-based Endeca, which promotes the concept of "guided navigation," presenting users with options to help them narrow their searches. Consumers also receive suggestions if the engine comes back with no matches.
"It's a very big improvement over what we internally have, which is probably 4-year-old technology," Kardesh said, noting that Spiegel expects to complete implementation of the product by Oct. 12. "The technology has gotten much better."
Free Insider Guide