Former Google Engineers Launch Search Engine

A startup founded by engineers from Google Inc. and other tech giants is launching a search engine that claims to cover three times as many Web pages as Google.

The startup, Cuil Inc., plans to launch its product Monday and aims to deliver better results than other major search engines by searching across more Web pages and studying them more accurately. The site’s results page resembles an online magazine — a different look and feel from search juggernaut Google’s.

Cuil

A Web-screen view from Cuil, whose results page offers a different look and feel from Google’s.

“You can’t be an alternative search engine and smaller,” said Anna Patterson, Cuil co-founder and president, and one of the engineers who helped build Google’s search index. “You have to be an alternative and bigger.”

Cuil, based in Menlo Park, Calif., is the most recent in a long string of search-engine startups to try to take on Google in an industry that has been difficult for even giants like Microsoft Corp. to crack.

Many have tried to compete by focusing on particular areas, such as searching images or allowing users to review and edit results. Many of these Google challengers have crumbled after failing to build enough scale to support their growth through advertising; a few others have been acquired by larger players.

Cuil has raised $33 million from venture-capital investors and has a deep bench of career search engineers, including Ms. Patterson and her co-founder and husband, Tom Costello. Mr. Costello built search technology for International Business Machines Corp. and was on the research faculty at Stanford University.

Greg Sterling, an Internet analyst with Sterling Market Intelligence, said that bench, along with the fact that the company has already built such a large search engine from scratch, bodes well for its ability to compete in the long term. But like all new search entrants, the company must still find a way to generate enough advertising revenue to fund the hefty infrastructure and technology costs of scaling a search engine, he warned. “It won’t be clear at least for a year or so whether they can break into the top group,” he said.

Ms. Patterson said other search startups have failed because they haven’t found a way to search more Web pages than Google.

Cuil, which claims to be able to search for results across 120 billion Web pages compared with across Google’s estimated 40 billion, says it has solved that problem. Ms. Patterson said it has developed a faster and better way to index Web pages that relies on fewer machines.

In addition to looking at the popularity of a Web page, Cuil also analyzes the concepts on the page and their relationships — grouping similar results under different menus. A Cuil search for “Bruce Springsteen,” for example, pulls up a section for results on the artist and a section for results pertaining to tickets.

A search on Google for “Bruce Springsteen” pulls up similar results — including the same homepage and some fan pages — but displays them in one long list of links. Google doesn’t comment on how many Web pages it searches, but a company spokeswoman said the search giant welcomes “competition that stimulates innovation and provides users with more choice.”

Cuil eventually plans to make money through advertising, although the service won’t display any ads at launch. Ms. Patterson said the company hasn’t decided whether to sell advertising itself or whether to partner with a third party. It has developed various mock-ups that allow users to collapse ads or to scroll through them, she said.

Cuil said it won’t collect personal information about its users, such as the addresses of their computers and their individual search histories — although it does track the terms people search for overall. While all major search engines have taken steps to cut back on the time they store data related to individual searchers and to make the data more anonymous, Ms. Patterson said Cuil can stop collecting information about individuals’ behavior altogether because its algorithms rely more heavily on analyzing the content of a particular Web page than on the popularity of the page.

Write to Jessica E. Vascellaro at jessica.vascellaro@wsj.com

Popularity: 1% [?]