One year ago, the search engine Cuil exploded on the launchpad. Hyped as a “Google-killer,” the site stumbled as its servers crashed and its algorithms spat out irrelevant search results.
Now, the Menlo Park, CA, startup hopes to stage a comeback in part by being the first search engine to pass search queries through users’ social networks to generate socially enhanced search results as a companion to regular ones. If, for example, a user searches for the band Green Day–and if she has allowed Cuil to access her Facebook account or any other social networking account–she’ll see a special box on the results page, showing those in her network who like Green Day and similar bands. The feature is expected to go live by the end of August.
“We are trying to leverage the information found on users social networks to enhance search results. This is similar to what Amazon or eBay already does: ‘People who bought this book, also liked this one,’ ” says Seval Oz Ozveren, Cuil’s vice president of finance and business development. “I think there has been a lot of buzz about this whole idea of social search, but nobody has actually done it to date.”
Of course, many social-networking sites already let users search within their networks. And other search engines are trying to expand into social networks. A search company called Worio, for example, offers a Facebook application that generates recommended Web links, akin to search results, based on analyses of the news feeds and other information from a user’s social network. If several of your friends live in or are talking about Miami, for example, Worio might provide Miami-centric links. The Cuil foray will be different: it will present the social network search returns aside the general ones.
It’s far from clear whether Cuil stands much chance of killing off any competitors with its move into social search, says Dan Weld, a computer scientist and search researcher at the University of Washington. However, he says, it’s well established that websites tied to trusted members of people’s social networks are more likely to be seen as particularly relevant.
“Statistics show that people, especially young people, are much, much more likely to click on a URL if they see it in a blog or Tweet from someone they trust. This clearly has a big impact on Web marketing and is leading a number of companies to develop tools to track and target these social influencers,” he says. “But whether it will really jive with search isn’t as clear to me.”
With $33 million in venture backing, Cuil was founded by Stanford computer scientist Tom Costello and a pair of Google alumni: Anna Patterson and Russell Power. Its core claim is that it searches more pages on the Web than anyone else–three times as many as Google. However, this hasn’t yet translated into comparable popularity.
To further distinguish itself, Cuil has recently begun offering other kinds of special categories of search returns along with the main ones. In March it introduced a “timeline”–a box on the right side of the page with search returns expressed by relevant date. A search for “Great Depression,” for example, brings up a box listing events of the late 1920s and 1930s–from various acts of Congress to the rise of Nazism–culled from pages that include dates. (Google has a prototype of a timeline search tool that can be customized; it produces temporally arranged links to news stories, from Wikipedia content or other sources.)
In June, Cuil also launched a “mapline”–search returns arranged on a map, and not just for obviously geographical searches like “pizzerias in Palo Alto.” For example, the “Great Depression” search produces pins on a world map; mousing over these pins (most of which are in North America) yields links to sites that, for example, describe Depression-era crop failures in Saskatchewan and 1930s public-works projects in Oregon.
But the question remains whether any of this will help resuscitate Cuil. According to the analytical firm Compete, shortly after its launch on July 28, 2008, Cuil had 2 million visitors–a figure that cratered to 130,000 by February and has stayed flat since then. However, Ozveren strongly disputes those numbers. She says Cuil’s traffic has been doubling every six weeks since February, though she declined to provide alternative numbers.
Explaining the disparity, Ozveren says that Compete does not accurately track “hover-over” activity–previewing a result without clicking through. Hover-over activity represents a growing share of Cuil’s traffic, she says. Also, when Cuil launched, it was the only search engine that didn’t track and store the Internet protocol addresses of its users’ computers. Ozveren argues that Cuil users may therefore be more privacy-minded than most, and therefore less likely to install Compete’s website-tracking toolbar.
Still, the numbers are far worse than the hype suggested was possible just one year ago. The foray into social search is one way Cuil is trying to recover.