In its second-largest acquisition, Amazon buys the company for $970 million.
(Page 2 of 3)
In addition, site search vendors are learning from other areas of the Internet as they develop their own products. FAST, for instance, looks to broader Internet search as it develops applications for users’ web sites. “The searches that people do at your site are tightly tied to the searches they do on the Internet at large,” Lichtman says. “The nomenclature they use to search the web at large may represent how they use your site.” FAST mines that data to help retailers understand how they should be returning results at their own sites.
Search vendors have also recognized that a close relationship exists between how site search results are displayed how Internet search results are displayed. SLI, for instance, displays results in such a way that search engine spiders can crawl the pages and find the data they need to populate consumers’ broader Internet searches. SLI released the Site Champion product at the end of 2005. That allows search results to be indexed so search engines can find the terms. “We’ve designed this with search engines in mind,” Ryan says. “It’s a fantastic way of driving more traffic to the site. Some customers are getting thousands of new referrals as a result.”
FAST also is offering a product that allows search engines to crawl a page. Based on search results at the site, FAST technology creates a page for the shopper to view the searched-for product. “And it’s a static page that Google can read,” Lichtman says.
Maintaining the core focus
But while site search has broadened its scope over the past few years, the focus of the core technology must remain on creating an outstanding shopper experience. And site search companies have not let that slip from their view. With most vendors having mastered basic search and navigation techniques, they are working on several developments that make today’s site search different from yesterday’s.
One of those is the ability to search web site content and databases apart from product information. SLI, for one, searches corporate policies, online reviews and other non-product information whenever a shopper launches a search at an SLI-powered site. Its system now can search up to 1 million documents at a site and will soon be able to search up to 10 million.
Another new development is automated normalization of data so terms that consumers enter can be matched to terms that manufacturers and retailers use to describe products. “We have our search solution integrated and compatible with a data transformation solution,” Leibow says. Mercado offers a product data optimizer that cleans up product information and makes it possible for like items to be displayed together in search results. “Most retailers work with multiple manufacturers and each manufacturer has a different way to describe the same thing,” he says. For instance, one bedding manufacturer will describe material as “cotton” while another will call it “CTN.” One will refer to “thread count” and the other to “TC.” “It all starts with the data,” Leibow says.
A further way in which site search has improved on the basics of returning usable results to shoppers is EasyAsk’s ability to do what Harris calls “category specific sorting.” That process allows retailers to set rules based on the category that a product falls into. Previously, all results would be displayed using a single criterion, such as popularity or date that the item was added to the product database. With category-specific sorting, one product-sweaters, say-could be sorted by popularity while another-shoes, perhaps-could be sorted by inventory levels. “It’s a major development,” Harris says. “For the search engine to know which category you’re in is not a trivial issue.”
Black is black
Site search vendors are also focusing on responsiveness and reliability. “Performance really, really, really matters,” says FAST’s Lichtman. “Site search is central to running your site and is a major part of the infrastructure. It should scale to the enterprise.” That’s especially true on e-retailing sites, where traffic is unpredictable. 30% growth in a year doesn’t mean 30% growth every day, Lichtman notes. A sale can drive traffic up by 60% one day and the end of the sale can mean only 10% growth the next. “You need something that can handle that,” he says.
Lichtman notes that FAST started as an Internet search company and continues to build on its expertise in handling searches of vast amounts of data in order to create site search that works. “The types of searches and number of searches that we are used to on the web translated very easily into the e-commerce world,” he says. “FAST is known as a leader in performance. Our customers spend their time thinking about how to drive traffic to their sites, not how to avoid it.”
Vendors are also pushing the technology into new ways of returning search results. EasyAsk, for instance, offers color-synchronized search results. At The Talbots Inc.’s JJill.com, for instance, shoppers who search on “black sweaters” get only black sweaters in their results. Those results are not easy to achieve, Harris says. It first of all requires taking a picture of every product in every color. Next it requires that the search engine recognize colors in search terms. Then it requires the technology to map to a particular image for display. “It’s subtle but very effective things that give you a pleasing response,” Harris says. “You say ‘black’ and Bang! the page is awash in black.”
With all the changes that the site search business has experienced in the past five years, it’s a safe bet that the evolution isn’t over. In fact, site search vendors are already working on the next generation of services.
One of those could be using site search to create personalized shopping experiences and even loyalty programs. EasyAsk, for instance, has been working with business-to-business sites and Harris expects that some of the techniques that the company developed for b2b customers will translate into business-to-consumer sites. For instance, search results at a b2b site have to take into account the contract that the business customer has negotiated with the seller, the types of products that the contract covers as well as the types of products that the customer has bought in the past. EasyAsk thus has to search not just product databases but customer history and contract information as well. The results are almost unique to each customer.