That seems like a surprising number. It’s hard to know for sure how accurate it is, or how effectively such reports might sell companies on Placed’s services, though they do help the company get media attention. It’s a bit ironic, though, because the analytics Placed offers its customers are a lot more sophisticated than clunky foot-traffic counts. For its new Placed Insights service, the company automatically collects data from people who have downloaded its smartphone app, so the population that it studies is self-selected, despite protestations to the contrary by CEO and founder David Shim. It’s not nearly as accurate as simply counting heads, as stores already do.

Nevertheless, the technology is both fascinating and promising — maybe even revolutionary. While big retailers and chain restaurants have developed sophisticated methods for measuring foot traffic within their stores, Placed also reports what customers do before they get there and after they leave. Further, it breaks down their demographic profiles and presents retailers and restaurant chains with the kinds of analytics that companies like Nielsen (NLSN) and comScore (SCOR) provide to online marketers.

MORE: Amazon’s (not so secret) war on taxes

A marketer, Shim says, might want to know: “Where do McDonald’s (MCD) customers go to eat when they’re not at McDonald’s?” If they tend to go to Panda Express or Burger King (BKW) rather than, say, Applebee’s, that’s a useful data point. (Placed doesn’t reveal the names of its customers, but Shim says they tend to be large retailers, chain restaurants, and ad agencies.) It’s even more useful to know the demographics of various customers: say, whether older, Hispanic customers of McDonald’s, as opposed to other groups, tend to prefer Panda Express over Applebee’s.

The ability to gather and manipulate this kind of data is unprecedented — in some ways, it’s even more useful and revealing than the data that can be gathered by tracking people’s online behaviors. Other companies provide somewhat similar services, but they gather their data (such as shopping habits) from third parties and cover wider geographical areas. Placed Insights gets it right from the source: the consumer, and the geography is measured almost down to the square foot. Placed Insights incorporates about 13 billion latitudinal/longitudinal geographical points and combines them with billions of other bits of data to produce any number of possible insights. This marriage of mobile technology and big data could revolutionize how marketers reach customers who are away from their computers, and shopping or dining.

Placed Insights uses GPS, Wi-Fi networks, cellular triangulation, and accelerometer and gyroscope technologies to determine not only whether a person has arrived at a destination, but whether they went inside, and even their movements in buildings.

MORE: Oculus Rift: Virtual reality 2.0

If all this sounds a bit Big Brotherish, it’s helpful to remember that all of the 70,000 people (at the moment) under Placed’s microscope have opted in to the service (in return for various incentives, such as small rewards or having money donated to charity). And Shim says that none of the data is attached to any user’s name when it is turned over to his customers. The company promises that users of its app will not be personally targeted with ads: The data is used only in aggregate form, for analytical purposes. The data is “normalized” using other data such as census reports. If a white app user happens to live in a mostly black neighborhood, for example, that data point is factored in — or out.

Not all that many people are interested in downloading Placed’s own app just to provide data for marketing purposes, so Placed strikes deals with app developers to put the functionality into other apps — often travel-oriented ones. Again, it’s all opt-in. Developers can actually make some decent coin from this: Shim says they get “a few hundred to a few thousand dollars a month,” which is particularly attractive since most of the apps in question are free.

While the group is self-selected (the app doesn’t really measure what the population at large does — it measures what smartphone users who have chosen to take part do), Shim says it is nevertheless as accurate as such a thing can possibly be. “People opt in to Nielsen ratings, too,” he points out.