With 70M places plotted, Gil Elbaz’s Factual has laid the foundation for the Web’s location layer

Gil Elbaz changed the face of online advertising by co-creating AdSense over 10 years ago. In his latest endeavor, he’s aiming to transform mobile advertising with Factual, a data science startup that operates a location platform used by companies like Apple, Samsung, Yelp, Bing and Tripadvisor.

Elbaz was one of the early contributors to LA’s ad tech AdSense scene by creating AdSense at Applied Semantics. When Google acquired the company, he stayed on to become Google’s presence in Santa Monica.

In 2007, Elbaz left Google to try something new, eventually launching Factual in 2009. The startup views itself as “a factory for data sets” with a current focus on data related to mobile and location.

Factual’s database contains location information for 70 million places and tracks at least 20 attributes per site. Clients include App.net, LivingSocial, Turn, Yelp, Bing, TripAdvisor, Samsung, Mopub, Gimbal, Groupon, Millennial Media, Trulia and Apple.

Last month, Factual released Geopulse Audience and Geopulse Proximity, products designed to help businesses better understand demographic data and employ fast, flexible geo-fencing solutions.

Factual isn’t solely centered around location. For instance, it maintains a product database for consumer goods found in grocery and drug stores. However, Elbaz believes that the “largest opportunity is to help mobile be an effective platform for distributing content, ads and a great user experience.”

“In the future, who knows, maybe we’ll use our technology to do different verticals but today we’re very committed to location and mobile,” Elbaz said.

Given how quickly mobile is growing, Factual already has its hands full. Mobile computing, by its very definition, takes users to new and varied places, so the need for fast and accurate location data is greater than ever before.

The original AdSense platform tracked the relationships among a couple million words. Factual’s data scientists are now trying to make sense of billions of data records.

The cost structure for computing has come down enough that the company is able to rebuild its entire system on a weekly basis. Elbaz estimated that the process of testing and training new machine-learning algorithms now costs $1,000, compared to $100,000 during his Applied Semantics days.

“We reprocess all our billions of unstructured raw data on a regular basis, which we couldn’t possibly have done before,” he said.

In addition to his role at Factual, Elbaz is also a strategic partner at TenOneTen Ventures. The firm is focused on investing in “great LA companies that are working on something ambitious, generally around applied data.” David Waxman serves as the managing partner. According to Elbaz, TenOneTen has invested in roughly a dozen startups so far this year, including SRCH2, Ranker and Kaleo.

Elbaz helped put LA on the map as far as global startup scenes are concerned, and now, with Factual, he’s well on his way toward transforming mapping itself.

Image credit: iStockphoto