• by Laurie Sullivan  @lauriesullivan, December 9, 2014

Reflektion has developed a visual site search tool for ecommerce Web sites resembling a combination of Google's Autocomplete, Shopping, and Product Listing Ads (PLAs) on Google.com. The tool, officially announced Tuesday, adds a layer of personalization and product images to site search results. Voice capabilities could come next.

The tool has been running in beta on Converse's, O'Neill's, and 2XU's Web sites. On some Web sites the results generate on average 315% more revenue per search term compared with standard site search, according to Sean Moran, CEO of Reflektion. "People who once use traditional search compared with visual search tend to buy about three times as much on the Web site," he said. "All the metrics increase, such as page views and time on site, but the startling number reflects the ability of consumer to more easily find what they look for, so they simply buy more."

It took O'Neill, one of Reflektion's initial customers, just a few days to install the platform, and saw conversion rates rise by 26%, with the average order size increasing by 17%, and page views by 85%.

Gander Mountain, AM Leonard, and Café Britt have also tested the technology. The idea to make visual and personal work see overall use of site search rise from between 5% and 10%, to between 15% and 30%, Moran said.

Put the cursor in the search box on the Web site and an overlay with trending categories and items appears. Type in a word like "blue" without hitting the "enter" key and items with the color blue appear in an ad box similar to PLAs in Shopping Campaigns. Click on the product in the results and the consumer connects to the landing page. The search is instant, visual and personal. It works with help from the company's transaction history and a product feed. A tracking beacon in Reflektion's technology on the site helps brands learn about their customers.

With Nike joining Reflektion's Series B investment round in January 2014, Moran feels confident the company's One-to-one personal touch for on-site search works. The technology analyzes the information that individual consumers share through their past and present engagements with a site, including historic searches, clicks, views, and purchases. That data is paired with machine learning algorithms that prioritize each visitor's search results to their individual characteristics as they type in real-time.

The next step could integrate voice into the platform. Reflektion launched in June 2013, supporting O'Neill with basic site search. Since then, the company went from supporting sites with hundreds of thousands of customers to millions of customers.