Location Accuracy: Three Common Questions

By
  • Facebook
  • Google Plus
  • Twitter
  • LinkedIn

dansilverData-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.

Today’s column is written by Dan Silver, director of marketing at xAd.

Things move quickly if you’re a mobile location technologist. A week can feel like three months, and if you’re not paying attention, it may seem like the world has passed you by.

That is also what it’s like for an agency media planner. That isn’t a knock on the planner, but a testament to their adaptability. The onus is on the technologist to make sure the media planner, agency or landscape, as a whole, understands the advantages they offer.

Among all the questions asked by agencies about mobile location accuracy, these are the most common:

1. What exactly is “lat/long” and how can it ever be inaccurate?

Let’s keep it simple. When a publisher sends an advertising request, whether it’s for an app or the mobile Web, a device’s location signal may be passed on using the phone’s built-in location services.

While there are many ways of deriving a device’s location, the most common are through global positioning system (GPS), geo-IP and registration data. GPS satellite-based data is the best-case scenario for advertisers, offering accuracy within 100 meters. Imagine the accuracy of your favorite navigation app, which uses the same data.

Publishers can also infer location by reverse geocoding the IP, but it’s a much less precise method with accuracy at the ZIP code level. Other times, publishers may relay a user’s self-inputted registration data, but the accuracy depends on how much you trust people, in general.

Whichever method of attaining this data, the publisher or ad network will convert this signal into a latitude and longitude coordinate. True GPS-generated lat/long data is often reported to be around 20% of all mobile inventory. The initial source of data will impact the accuracy of the lat/long reading.

2. Why does location accuracy matter if I’m running a branding campaign?

Location accuracy is even more important in a branding campaign because scale is key and any ad waste can greatly reduce ad effectiveness. With a branding campaign, you want to serve the most effective advertising at the most efficient cost. Therefore, if you are serving wasted impressions, you are greatly reducing the effectiveness of your buy. Location data passed through devices vary on a number of factors, and up to 80% of mobile location data are inaccurate.

So why would an impression be considered wasted? Registration data is an easy example to use. Many apps serve location signals based on user-generated ZIP code registration data. Think about college students who use their hometown ZIP codes for apps. They may be targeted for certain campaigns, when in reality they could be hundreds of miles away, out of state, or even across the country attending school.  The most common ZIP code used for registration is 00000. You can imagine this would not be very helpful when trying to determine a consumer’s current location.

3. My client doesn’t have physical store locations, so why would I need location-based targeting?

Location-based targeting isn’t only about serving an ad to a desired location.  Through accurate location intelligence, we can define and build real-world audience profiles.

Let’s say you are promoting a new golf shoe. One way you might want to target your audience would be content-based, such as running advertisements on golf or sports-related websites or apps. This makes sense, but how would you really know you are reaching actual golfers, as opposed to casual sports fans?

How about running ads that are reaching consumers who are physically on, or have been on, golf courses. You could also target consumers who are in, or have been in, sporting good locations – true in-market athletes. How about targeting consumers who are searching for the nearest golf facility on their mobile device?  Consumer behavior is an incredibly powerful way to define audiences, so consider this the next time you question the best way to reach your target market.

The accuracy of location signals for mobile advertising can make a world of difference for your targeting needs. Having a better understanding of the actual technologies that are being deployed in today’s marketplace will make you a smarter advertiser. Don’t be afraid to ask questions – it’s a technologist’s job not only to innovate, but to educate.

Follow Dan Silver (@danjsilver), xAd Inc. (@xAdInc) and AdExchanger (@adexchanger) on Twitter.

  • Facebook
  • Google Plus
  • Twitter
  • LinkedIn

Email This Post Email This Post

By on at

2 Responses to “Location Accuracy: Three Common Questions”


  1. Of the most common ways to derive a device’s location, you mention that GPS is best. To clarify, GPS comes closest to narrowing down an exact location, but it isn't necessarily best at targeting your ideal audience. The type of location-determining method that a brand deems “best” depends on the context of its ad. GPS is best for ads that require a user’s current location; think of a window-soundproofing manufacturer that wants to serve its ads to companies located near train tracks and airports. By accessing GPS information, the brand is able to determine the mobile devices that reside near highways and airports from 9AM to 5PM and geotarget them. Next, geo-IP or reverse geocoding is best for campaigns that target an area of interest, such as a city. Let’s say that a restaurant franchise wants to serve an ad for a local location, targeting residents in and visitors to that zipcode as well as surrounding ones. Reverse geo-coding is the best option for this. Lastly, campaigns that target home location might opt toward registration data. If a medical network wants to advertise a local office to people whose permanent residence is in that town or any bordering zipcodes, then the network might use device registration data. As you mention, with all of these, you need to factor out the waste, like 00000 zipcodes.

  2. Elise, I have to disagree with these comments. Implying that less accurate targeting is "best" for a situation seems inaccurate. The problem with geo-IP targeting is that it is less accurate at best, just plain wrong at worst. Why would that be better for city targeting than actual GPS data from the device about whether it is in the city? It seems similar assuming the IP is not dynamic and has not hopped to another city.

    Similarly, registration data is probably the most likely to be just plain wrong here - as you point out, much of the registration data is simply inaccurate. Also, registration data does a poor job of identifying people that move frequently such as Generation Y and college age audiences. Longitudinal models of GPS data are probably the most accurate way to identify household locations.

    While there are obviously location-targeted campaigns that don't require high fidelity GPS, the situations where other data yields better results are probably edge cases.

Leave a Reply