As anyone who works in marketing knows, location is one of the most important elements you can use to target an ad campaign. In traditional media, most geotargeting is implicit; if you buy ads in the San Francisco Chronicle, you can feel confident that only people in San Francisco are going to see that ad.
In digital however, you can’t assume; YouTube and Facebook are just as easily accessed from Japan as they are from New York. But beyond controlling cost and ensuring ads only run within the footprint of where the advertiser actually sells their product, geotargeting offers huge opportunities for marketers.
For one, geotargeting in digital allows for more sophisticated measurement and personalization than was ever possible with traditional media. And with the proliferation of mobile devices, and the remarkable granularity and specificity the provide in terms of location, geotargeting has never been more powerful.
But while many marketers understand the value of geotargeting, not many are likely to understand how the technology behind it works. Having a firm grasp on the technology though is actually critical in this case, as different solutions take different approaches to the problem of determining the physical location of a consumer, and the simplest solution is usually the least accurate. Just as traditional strategies don’t always translate well to digital, many desktop strategies don’t translate well to mobile.
In online environments, ad servers look at a user’s IP address to figure out their location. Behind the scenes, the ad server maintains a large database that has every IP address already mapped to its country, state, and postal code. So, when a request comes in, the ad server strips the IP address from the header of the request, queries this table, finds the necessary location data, and then picks an ad that matches that criteria.
Now the ad servers don’t create this table themselves, they license it from another company like MaxMind or DigitalEnvoy, whose primary business is geolocation data. This is no enviable task; IP addresses themselves don’t necessarily have an obvious pattern in the way they are assigned like a telephone area code would. It’s a bit like solving a mystery, and the geolocation companies use a variety of methods to approach the problem.
The first thing to do is figure out which ISPs own what IP ranges. This is public information, used so that an IP isn’t shared among many users across different ISPs. And since ISPs tend to serve a particular region, usually a country can be assigned to the user with this information alone.
Now, to figure out state and postal code involves a more complex process. At the core though, the geolocation services build up a network of servers from which they can send out pings, or connection requests, and known physical locations of public entities like universities and government office IPs. Eventually, with enough data, the geolocation company has the capability to triangulate any IP on the web.
It works like this – if there is an IP address the company wants to locate, they ping it from a few of their servers, for which they already know the location. A ping is just a way to test if a computer can connect, and how long it takes to do so, but doesn’t transmit any meaningful data. Then, by looking at the time it takes each server to connect, it can establish a shared point or origin, and thereby physically locate the user. It uses the public IP locations to validate their approach and check for anomalies in network latency which would lead to bad data.
The risk to this approach is that it isn’t always terribly accurate beyond the city to zip code level. If, for example, you were to use MaxMind’s demo service to locate your own IP, it will likely show you perhaps a mile away from your actual address, likely at the nearest network node, the point at which your computer connects to your ISP’s network infrastructure.
For this reason, some companies have taken a more direct measurement approach to IP geolocation vs. trying to infer it through ping triangulation. It’s far more straightforward, but requires a lot more manual effort. Basically, these companies send cars out to drive up and down every street in the country and log WiFi IP addresses as well as their physical location to populate the same table that more traditional geolocation companies build through technical means. Google and Skyhook both use this approach.
By utilizing a premium proxy service, online professionals can view, monitor and validate localized geo-targeted web/mobile content and ads as it appears in each geolocation.