Factual Wants to Put the Web in OrderBy
(This story was corrected to show that Factual raised $2.1 million this year, not $15 million).
Gil Elbaz recalls meeting with a prominent venture capitalist during the late 1990s to discuss his first startup, Applied Semantics. When Elbaz said the company could someday generate $100 billion in sales, the potential investor was dismissive. "He about threw me out of the room," says Elbaz. In 2003, Google (GOOG) bought Applied Semantics for $102 million. It became the foundation for the search giant's AdSense, which places relevant text ads on Web pages and in 2009 accounted for 30 percent of the company's $24 billion in revenue.
Elbaz commands more respect these days. Earlier this year a who's who of Silicon Valley power players invested $2.1 million in Factual, the company Elbaz formed after leaving Google in 2007. Factual, which opened its service to the public in October, is organizing bits of useful data strewn across the Internet. Every day, names, numbers, dates, addresses, and other pieces of what Elbaz calls "unstructured data" are added to the Web. Since it doesn't follow any standard formatting, this data is useless to most computers. Machines can't readily grasp concepts such as "middle school" or "junior high," let alone distinguish how they're related.
Factual aims to automatically find and sort this data, amassing a comprehensive collection of datasets that businesses will pay to incorporate into their websites and mobile applications. Inside Factual's Santa Monica (Calif.) office, a few miles from the old Applied Semantics facility that became Google's Southern California outpost, a couple dozen engineers train computers using principles from an academic field called ontology, which is the science of categorizing words and phrases. The algorithms are smart enough to know that junior high refers to 7th and 8th grade while middle school includes 6th grade as well. They're also smart enough to update the database when online information changes and verify its accuracy. "This is a very tough problem to solve well, even for Gil," says Elon Musk, the PayPal (EBAY) co-founder who went on to create Tesla Motors. "But it would be unwise to bet against him."
Factual is building databases on health, entertainment, and other topics, but for now its primary focus is on basic information about businesses and geographic points of interest. Individuals and small developers can use the datasets for free, but larger companies must pay if they want to integrate Factual's information into their own products. On Nov. 4, Facebook said it started using Factual's service to populate Facebook Places—a feature for its smartphone app that lets users broadcast their location to friends—with the names and addresses of points of interest in the U.K. and Japan. San Francisco startup Booyah plugs Factual's data about local businesses into MyTown, a mobile game that involves players visiting real-world bars, restaurants, and other hot spots. Navigating Cancer, a website that helps recently diagnosed cancer patients, displays a Factual-powered list of oncologists, including details on what type of insurance they accept and areas of specialty.
Factual has yet to turn a profit. The company charges businesses that use its data 50 cents for every thousand data requests. On average, such customers request about a million pieces of data a month, totaling $500. Some customers, including Booyah, get discounts for helping to keep Factual's information fresh by, say, verifying whether a business is still open.
The key to expanding the company is persuading software makers to farm out the work of data management, says Eva Ho, Factual's vice-president for marketing. "Their core competency isn't around getting data, organizing data, and keeping it fresh," she says. "Their core competency is around building a killer application. They should outsource the whole data management service to somebody like us to deal with the heavy lifting."
The area of data-as-a-service gained prominence in July, when Google bought Metaweb, a rival to Factual, but one that relies more on human editors to create and update its trove of data. Recently, executives at other Internet giants, including Yahoo! (YHOO), have expressed interest in building or acquiring data processing tools of their own. While Elbaz won't rule out a sale, he says for now he's focused on amassing data. "Would it be fun to get an acquisition offer from [Google] again?" Elbaz asks. "It would be flattering."
The bottom line: Factual organizes the Web's messiest data and sells it to corporate customers such as Facebook and Booyah.