Yesterday Stefanine Olsen reported from the Web 2.0 conference about a new project from Idealabs's, Bill Gross. Details in this blog post.
"It's going to be controversial, but it's awesome," Gross said in an interview at the Web 2.0 conference on Tuesday. "We're trying to improve search productivity.
It's also one of the first search engines to harness data on "user intentions," extrapolating meaning from words typed into a search box. It's done so by licensing data feeds from third-party Internet service providers, which have tracked, anonymously, what people do after they've typed in a specific search terms. It uses this special sauce, a data feed of more than a terabyte, to compute the relevancy of certain searches and resort results.
"...the most "subversive" part, Gross said, is that the service is transparent to advertisers and visitors. Snap will make money by selling advertising placements at the top of search results, but the twist is that it will let marketers pay specifically for people who buy at their site as a result of the Snap listing, a "cost per transaction" model. Furthermore, advertisers can specify, for example, that they want to pay 25 percent of their product cost or $4 for every widget they sell if a consumer buys it from the ad at Snap--and that information will be displayed in Snap's product listings. In comes the transparency.
Snap has licensed search technology from desktop search player x1. Data comes from Gigablast, Looksmart, along with "anonymous data feeds" from ISPs.
He writes, "We even are revealing OUR REVENUES. Our conviction is that you get better results because transparency prevents advertisers or others from gaming the system.:
The Snap help section has a bit more info about their ranking algorithm. Say hello to Snap Rank!
"...'Snap Rank' algorithm combines advanced calculations involving analyses of the web itself (such as link strength and query matches) with a large body of human data to deliver results that are accurate, trustworthy, and directly related to your intentions and needs. Snap processes post-search actions from our Snap Network, started in January of 2004. Snap evaluates where users go after a search - what listings they choose and how many pages they review on that site, right after the same search you did. The overall popularity of the site also influences the ranking.