We should probably define "major". If your site is visited by tons of mobile visitors and you use the geo functionality to impact user experience, then for you

We should probably define "major". If your site is visited by tons of mobile visitors and you use the geo functionality to impact user experience, then for you, it may be. However, most sites continue to have far more desktop usage than mobile, on average. So, for most visitors to most sites, it's not really all that major. YMMV, of course. In fact, GeoIP has suffered from the problem you describe for a decade now, from every AOL user appearing to come from Virginia, USA, all the way to modern corporate proxy systems which have vast numbers of visitors all appearing coming out of, say, White Plains, NY, USA (yes, there are lots of corps there... but not every employee works in that location). So, your idea is a good one for mobile where we kind of know the big "offenders" (Kindle Fire may rise up to become an issue as well), but it's hard to figure this out for every IP. Commercial GeoIP companies often offer a "confidence value" with their data which reflects how confident they are than an IP would really reflect the geo of the user; unclear if this service will have that. In addition, a good user experience might be to personalize the session but allow user feedback, ala "we think you are currently in Antarctica; click here to tell us where you really are..." You might also decide not to use some functionality in your "mobile skin" or other version of the site, if you do separate designs for mobile vs. desktop. I say, best of luck to them. Web analytics continues to suffer from a disconnect: vast reporting on users, but little connection to taking that data and impacting a user's visit, even in future visits, much less in the same session. Let's see some folks push this a bit, even if the data has some problems (which, let's face it, is always a problem with any data involving the web, imho).