• by Eli Portnoy , October 18, 2013

If 2012 was the year of mobile advertising, than 2013 is certainly the year of location-enabled mobile advertising. As the industry has evolved marketers have come to understand that mobile is not a tiny PC, but rather a unique medium with its own set of best practices. The cookie-based user profiling of desktop doesn't quite translate to mobile, but advertisers have recognized that equally relevant analysis can be made based on a user’s location. In a sense, location has become the cookie of mobile; however, for advertisers to effectively use it, the location must be accurate. 

The problem is that the location data most advertisers are using is wrong.

Thinknear ran a test in which we served over half a billion ads, purchased through some of the largest inventory providers in the industry.  Every impression we served had a specific location associated with it. To validate the location of these impressions, we asked users who clicked on our banner to allow us to pull their actual location. We compared this actual location to the location provided to us by the inventory source. The results were eye-opening:

Twenty-six percent of reported locations were off by over 10,000 meters, while less than 33% were accurate within 100 meters. To put this in context, if a luxury car dealer wanted to target high-end shoppers on 5th Ave. in Manhattan, 1 in 4 of their ads could unintentionally show up in residential New Jersey. Not only that, but more than 42% of impressions were off by at least 3,500 meters. So in this example, the data showed that the luxury shopper is just as likely to be an NYU college student playing "Angry Birds" at Washington Square Park as they are to actually be a 5th Avenue shopper.

To understand the discrepancy, we have to start with the source of the data. Publishers and inventory sources have three ways of generating location information. In order of decreasing accuracy:

1)     They can pull accurate location using the phone's location services (GPS or WiFi triangulation)

2)     They can infer location by reverse geo-coding a user's IP address

3)     They can ask for and rely on the accuracy of a user's registration data.

The precision of each of these methods varies dramatically. While GPS and WiFi pull accurate location, registration data is often outdated and completely inaccurate. However, regardless of the way in which this information is generated, publishers and ad networks convert the location into a latitude and longitude. This means that advertisers may unknowingly be treating impressions as accurate to within a few meters, when -- in the worst case -- they can be off by thousands of miles.

Not only does this inaccurate location affect demographic targeting, but it can have serious UX implications as well. For example, if a QSR decides to target users within 1 mile of their restaurant with a banner that reads “Tasty Burgers Only 1 Mile Away!” and provides driving directions to the restaurant, then they have created a highly engaging user experience (and potentially gained a new customer). However, if a user clicks for directions and realizes that they are in fact 20 miles away from the restaurant, the experience is compromised and the effect of the campaign is diminished, both from a branding and a lead-generating standpoint.  

Mobile’s trackable, on-the-go nature sets it apart from any other advertising medium. Not only does a user’s location tell their story, but it can influence their likelihood to convert. For example, in one study Thinknear found that when within a 5 mile radius, 4x as many users drove to quick service restaurants when compared to those within a 5 to 10-mile radius. The ability to leverage this data is what will likely drive mobile into the forefront of digital advertising. The value is clear, but there is much room for improvement. Better location data will lead to both improved campaign performance and user experience. So as location-based targeting becomes a bigger piece of the mobile advertising pie, advertisers and buyers must be diligent to filter out the lat/long from the lat/wrong.