Hey, mobile developers, have you ever wondered where users are when they interact with your apps — as in down to the level of whether they’re in a Starbucks or the McDonald’s right across the street? A Seattle-based startup called Placed has a novel approach to mobile-device data that it claims can tell you just that, a capability the company thinks can change the way developers think about everything from targeted advertising to product design.
Placed Founder and CEO David Shim says the company wants to change the discussion around mobile analytics away from metrics such as how long an app was open or someone’s basic location for simple geo-targeting. The question Placed wants to answer, he told me during a recent call, is “How does context of place impact what people do within a mobile device?” Thanks to a database of more than 300 million locations, Placed users can slice and dice data by geography, time of day, business name, type of business — 400 criteria total — to determine where and when their users are engaged with their apps.
This data could be especially compelling for maximizing ad revenue, of course, by providing valuable information about where potential consumers are in relation to a given advertiser’s products. If you have a travel-booking app, knowing that people are booking hotels while sitting in train stations might be useful, he explained. If people are scanning bar codes with your app, you (and potential advertisers) might want to know whether they’re doing that most often in Costco or in Best Buy. You might just want to know your audience is “more Whole Foods than Albertson’s,” Shim said, or that users are most often nearby a Subway when they’re using the app.
Knowing more about where your users are could make a difference on the product-development front, too. One private beta customer, he said, saw that people were using its app a lot while driving and decided to add voice controls as a result. Placed’s analytics dashboard also lets customers track movement — they can, for example, watch app users collectively move from the suburbs to the city and back during the course of a day. For a restaurant-rating app, perhaps that means it’s time to incorporate recipe reviews for users who usually end up cooking dinner rather than eating out.
In the demo Shim gave me, it was easy to see differences geographically (some of which are clear in the infographic) as well as by time of day (although Subway and McDonald’s are close in terms of overall percentage, people are in or near them at entirely different times). And that people in Atlanta use apps while driving a lot more than do people in Seattle.
Millions of data points make it happen
If you’re wondering what makes Placed different than just tracking the GPS signal within a user’s phone (and which many apps already have permission to use), Shim said it’s a matter of granularity. Its model extracts signals from noisy location data in a fairly unique way to show where someone is, right down to the very business.
For example, while a GPS signal might be very accurate in the parking lot, it can become muddled when someone enters a building. All of a sudden, someone clearly in the Costco parking lot is in a half-square-mile area that could just as easily encompass the McDonald’s down the street. Because of Placed’s location database — of which the company has already verified more than 600,000 locations — it’s able to infer that the user actually entered the Costco located at that very spot.
Or, in another scenario, Shim explained, Placed’s model can determine a user’s whereabouts by process of elimination. If every other business in an area is closed, there’s a good chance your user is in the sports bar that’s still open. As noted above, the model also takes into account factors to infer whether someone is walking or driving while using an app.
It sounds simple enough, but it’s not. In order to accurately infer where people are, Shim said, you have to be able to stitch together a series of data points in a meaningful manner. The company already analyzes 24 million raw locations a day in its Hadoop cluster in order to add structure to messy data. Placed has ties to former Seattle big-data startup Farecast (now part of Microsoft), but “we collect more data in a 30-day period than Farecast had total,” Shim said.
Placed also has four Ph.Ds. on staff constantly working to refine its model by taking into account new data points such as captured images or search queries that might indicate where someone is, and specific mobile devices’ varying accuracy of location data.
However, any discussion about a cool new approach for tracking users’ locations is bound to bring up privacy concerns. When I broached the subject with Shim, he was quick to point out that Placed takes privacy very seriously. It only works with apps that already have permission to access users’ location data, and Placed only delivers aggregate analytics.
Because it doesn’t let its users see individual results, he said, “someone’s behavior being measured isn’t going to affect them at all [in the form of individually targeted advertising or the like].” Rather, a company like McDonald’s can confidently place ads on a given app knowing that a certain percentage of its users will be within spitting distance of a McDonald’s (or a competitor) a large amount of time that app is open, presumably making them more willing to actually pop in for a hamburger.
I think Placed is onto something with its service (which is currently free, by the way) and hints at what we can expect as that mobile analytics space begins to heat up in the next year with a focus on developers rather than just marketers. Along with companies such as Structure Launchpad winner Keen.io and some stealth-mode companies ready to hit the public eye, mobile developers — many without access to corporate IT resources or analytics tools — are going to start getting the data they need to build better apps and to put the power to monetize apps into their hands.