This is the third article in a four part series on Geotargeting. Read parts one and two.
Despite the complexity and scientific approach of IP based geolocation identification, there are well known limitations and inaccuracies with the current methodology. While geolocation data is usually extremely accurate down to the state or city level, as services demand more granular data, many of the current geolocation services start to break down. The loss in overall coverage is quite small, but accuracy can be another story.
Server Location vs. Machine Location
One of the more challenging aspects of IP based geolocation is that often times, geolocation services end up using the location of the server on which that IP is accessing the internet, not necessarily the location of the end user’s machine. So however impressive you may have found the diagram in the last article on IP triangulation, the method may end up targeting the wrong location. The classic example of this known within Ad Ops circles was AOL dial up service, which in its heyday represented a large share of internet users. AOL’s servers were all physically located near its headquarters in Virginia, so every IP address hosted by AOL, was often shown to be located in Virginia, even though users were spread throughout the country. Today, this is much less of a problem because most consumers have a high speed connection serviced by a locally hosted ISP, but it exposed the problem in a big way at the time.
That said, local ISPs network routers, while usually quite close, are frequently in different zip codes, so while coverage remains high for most IPs at a granular level, accuracy can be less reliable. When researching this article from my location in New York City, most services were more than 7 miles off my actual physical location, perhaps a small difference in much of the country, but an enormous gulf in as dense an area as Manhattan. Every service however was correct about my location at a country, state, and city level. You can check your own location on MaxMind’s demo page, which incidentally, was one of the more accurate services.
Source Data Corruption
The WHOIS directly also has some notable downsides – registrants aren’t forced to keep their information current with their domain registrar, so the information can quickly lose accuracy as businesses move, open new locations, or even sell to a different owner.
QA Geotargeting with Proxy Servers
Finally, it is fairly simple to manipulate your IP address to seemingly change the location of your device online. Using a proxy server, for example, any user can fool geotargeting services with minimal effort. Proxy servers simply create a middleman in the network routing process, a sort of forced detour through a particular server in a particular location, which redirects the information to the end user in a separate process, which is kept hidden from the information provider. If you have access to a VPN service at your workplace, you are using one form of a proxy service.
The service is actually widely commercialized, and quite useful to Ad Ops as a QA tool, and is available through companies such as GeoSurf, GeoEdge, FoxyProxy, SquidProxies, and many others. There are free options available as well, though they don’t work quite as well and can be loaded down with spyware, so be cautious in trying them.