SAN FRANCISCO — A new company launched by dotcom survivor idealab aims to take a chunk out of the search market by letting users slice and dice their search results.
Snap, which launched in beta on Tuesday, is an attempt to combine natural search with revenue-producing activities including comparison shopping and paid listings. The service was created by Perfect Market Technologies, a company developed by high-tech incubator idealab of Santa Monica, Calif.
The types of search results generated by Snap are context-sensitive; what information is delivered depends on the search term. For example, if the search term is in a category where structured data is available, such as “camera,” results include a table with brands and models of cameras along with product attributes, thanks to Smarter.com, another idealab company. The user can refine the search by any of the attributes, from brand to model to resolution.
The company also touts “transparency,” offering advertisers Overture-style pay-for-placement listings while showing searchers exactly how much they paid. It’s a method premiered by idealab’s most successful company, GoTo.com, which became Overture Services. Overture was sold to Yahoo for $1.6 billion in 2003.
“We’re exposing every action and transaction to avoid dead ends in searches,” idealab founder Bill Gross told the audience at the Web 2.0 tech conference here, where he unveiled the technology.
“Snap Rank,” an algorithm that looks at user satisfaction, site popularity data, and the amount advertisers pay, sorts the other search results. Paid listings are identified with a “paid result” label. Gross said the greatest factor weighed in the ranking process is user satisfaction. User satisfaction is, in turn, most heavily influenced by the conversion rate — a measurement of sales, downloads, or leads generated as a result of clickthroughs to a site.
The Web site popularity information is gleaned from clickstream data from a network of 1 million users whose online movements have been tracked anonymously since January. This data comes from log files licensed from a few large ISPs, with all personally identifiable information removed. Snap.com processes the several terabytes of data in the files to identify patterns of user behavior after a search.
The popularity of a given listing, the site it links to and the number of pages viewed by visitors once they arrive there, are all factors that affect search result rankings. The searcher can sort or filter the results by one or more of these values using technology licensed from Idealab’s X1 Technologies.
They also can refine the search by typing additional terms into a special box. The search engine automatically responds by narrowing the results without the necessity of loading a new page.
“If someone completes a transaction, that’s the ultimate proxy for a satisfied searcher,” Gross said. Because search engine marketing techniques have made it relatively easy for sophisticated advertisers to get high rankings in search results, Gross explained, it’s not always a given that the top listing will be the most relevant for the searcher. Factoring in the top action people took after the left the search results greatly increases the relevance, he explained.
Gross and his team are hoping to make Snap one of their more profitable ventures. Idealab nurtured Overture’s patented method of ranking paid listings. Search rival Google then paid Overture around $280 million in stock to settle a suit claiming Google AdWords infringed on its patent. But idealab also had its share of dogs, including Pet.Net and eToys.com. Meanwhile, Snap! was the name of a failed portal and search play by CNET Networks.
Editor’s note: ClickZ writer Kevin Newcomb contributed to this story.
