Magazine Article | October 18, 2006

Intelligent Web Site Searches

Source: Innovative Retail Technologies

By implementing site search technology, a coffee retailer amplified the number of searches performed on its site by 50%.

Integrated Solutions For Retailers, November 2006

Displaying accurate results for a search performed on your Web site can determine whether consumers purchase products through your site or your competitor’s. Statistics revealing consumers’ actions based on search results on your site are imperative — if measurements don’t exist, it’s tricky to identify problems with site searches. Green Mountain Coffee Roasters, Inc.’s Web Architect, Doug Cadmus, found functional and statistical deficiencies with its site search tool following a platform change to Microsoft Commerce Server. The retailer sells coffee and gifts to grocery wholesale customers (e.g. restaurants, hotels) and directly to consumers via catalogs and its Web site.

“Internet users are influenced by the simplicity and power of search engines like Google,” says Cadmus. “This makes search functionality on retail sites essential to uniting consumers with sought-after products. If search results are generic, they are only as effective as random browsing.” After the platform switch, Green Mountain’s Web site lacked the functionality that Cadmus wanted, such as sorting search result sets, identifying misspellings, and indicating items currently out of stock. As a result, Green Mountain representatives visited the Internet Retailer conference to find a new Web search solution. After comparing products and considering its employees’ prior experience with search products, the retailer chose SLI Systems Learning Search (a hosted and managed site search tool that learns from visitors’ click behavior), Search Suggestions (insight into tricky keywords and misspellings), and Site Champion (automated search engine optimization).

“With its Learning Search system that constantly refines accuracy, the solution from SLI Systems surpassed competitive software,” says Cadmus. “When searches are performed on our site, a component of the search engine compares the keywords and result sets to past searches, then compares those with the merchandise that users select from within each result set. As users click on links within the result sets, those products become more prominent over time, enabling the search tool to react to consumer behavior. It does so by identifying relevant result sets based on keywords and prioritizing dominant merchandise toward the top of each result set.” The system also makes suggestions to keywords used in searches, such as spelling variations, to ensure accurate search results.

Update Available Merchandise, Increase Site Search Use

Green Mountain regularly publishes catalog and Web site updates, as merchandise availability varies. So, timely updates to product availability are required for accurate search result sets. Green Mountain exports an XML file from its catalog database that SLI pulls every 15 minutes to feed search sets. This file also designates out-of-stocks, so it essentially tracks inventory, too.

“I receive a weekly report via e-mail from SLI displaying aggregate search information,” says Cadmus. “It includes data such as the total number of searches completed on our site, the clicks within result sets, and the changes in result set sorts [e.g. display by price or category]. I can drill down for quantifiable results displaying an increase or decrease in searches for specific coffee flavors or products. I can also follow a path of keyword relevance and gauge the accuracy of search results.” When Cadmus studies consecutive reports, he identifies increased accuracy of keyword searches as the software shifts the order of merchandise within result sets.

The retailer is certain of one thing — consumers now use the search function more frequently. Immediately following the installation, approximately 600 searches per day were completed on the retailer’s site. Today, approximately 1,300 searches are performed each day. Overall, Green Mountain’s sales have increased significantly within the last year. Cadmus cannot say the growth is due only to this software, but he does believe the enhanced search is one factor. Because of the increase in searches completed and the criteria tracked on the weekly reports, Cadmus surmises that returning visitors find the site’s search a valuable tool. As a result, he recently redesigned Green Mountain’s Web site to make the search field larger and more prominent, enabling visitors to enter more detailed search queries.


 For More Information On SLI Systems
Go To