A Trillion Points of Data

The idea is no pipe dream. TomTom, the Dutch manufacturer of GPS devices, has been using cell-phone data to provide up-to-the-second traffic information in Western Europe since late 2007. "We're able to have almost complete visibility" of the roads, says Tom Murray, the company's vice president of market development. "It's a game-changing, killer application that rounds out our portfolio of navigation tools." Like Nokia, last year TomTom completed its own purchase of a major digital mapmaker, spending $4.2 billion to snap up Teleatlas. Nokia, for its part, is investing in a cell-phone-based traffic system, starting in Northern California. It has partnered with the University of California, Berkeley, to conduct a field test involving 10,000 cell-phone users, and will eventually sync their calendars and commuting habits with current traffic flows. In practice, this means your phone can alert you when you're running late.

Drivers carrying cell phones are creating a wealth of data about average traffic conditions, allowing Eric Horvitz, a computer scientist with Microsoft's research department, to concoct something he calls "surprise modeling." There's no point in being told that Highway 101 in San Francisco is backed up at 6:30 on Friday—the 101 is always backed up at 6:30 on Friday. The useful information is when the road is uncharacteristically crowded, or remarkably clear. "The idea is to harness [these datasets] to let me know in advance of a surprise coming," he says. Inrix, spun off from Microsoft in 2004, provides this kind of information to navigation companies and mapping programs.

Cell phones potentially give us a window into more interesting human patterns than traffic. Having a bird's-eye view of where people are and where they're going, in real time, could be a rich vein of information on behavioral trends. Sense Networks, a three-year-old startup housed in a hip SoHo loft in downtown Manhattan, has 15 researchers (including 5 Ph.D.s) trying to figure out what's possible to glean from location data. The company's goal, says CEO Greg Skibiski, is to index the real world much as Google has indexed the virtual one. "Google looks at the context of a Web page by the pages that link to that page, and the pages that link out from that page," he says. As a result, they "have a good understanding of the whole index. We do the same thing using places and location data."

Too abstract? Here's a real-world example: Say a jazz group plays a 10 p.m. set at a downtown bar. Using the location data they've collected, Skibiski and his researchers can see where all the jazz aficionados ate dinner before the show, and what kind of late-night clubs they visit after the trumpets hit the final high C. They're putting the jazz club—and, by extension, its patrons—in the context of the rest of the city. That capability is on display in the company's first application for consumers, CitySense, which shows where everyone is—in real time. Cell-phone users who download it can see which blocks are busier than usual, and even learn the most popular destinations people go to from their current location.

That might not sound like much, but if you're a business owner, it's sweet music. Businesses spend enormous amounts of time and money trying to understand their customers. Surveys and focus groups, though, are blunt instruments. Sense Networks can craft customer profiles based on where people actually go and what they actually do—not where they say they go and what they say they do. "We're seeing the real, unbiased you," says Skibiski. "And we're removing the bias error without asking you any questions." Consumers, theoretically, should be pleased, too, because businesses will be able to offer targeted services, such as customized Web searches.

Sense Networks has raised $3 million from financiers like Steven Drobny. As Sense began developing its algorithms, the moneymen saw an opportunity to use them for stock-picking. "We can use [our models] to predict retail demand," says Skibiski. "How many people are coming to Macy's versus Saks versus Nordstrom?" Since mid-2006, the firm has used location data from San Francisco to operate a trading portfolio. "It's in the green," says Skibiski, "so something's working."

The tie between location data and economic activity may even be strong enough to predict imminent financial crises. The company's analytic software charts various consumer habits, culled from location data, against trends in finance. One chart plots San Francisco nightlife patterns on the same graph as the Dow. It shows that, just before the market peak in July 2008, urban partygoers hit the town at later-than-ever times. Velvet-rope lines at nightclubs might just be the new leading indicator of financial panics.