While readers of this blog and others like it are sophisticated about the importance of and opportunity tied to location (i.e., offline buying), the rest of the world . . . not so much.
Here’s an excerpt from a MediaPost interview of Quova’s VP Kerry Langstaff explaining the importance of “knowing where the user is”:
BI: Can you give some examples of how a publisher or advertiser can custom-target consumers based on geo-location?
Langstaff: So, as an example, say a customer is searching for shoes. By using IP geo-location data to situate exactly where they are, a shoe retailer can localize its landing page for each incoming visitor. Another customer example is a newspaper which uses reader location to customize their news and ad content. If you log in from Massachusetts, you’ll get the Red Sox score first — not the Yankees score. And news, weather, and the store locations of advertisers can be localized based on where a particular person is logged in from.
This particular publisher has 60 DMAs they serve, from Boston to San Francisco. Not only can they geo-localize targeted content, but they are able to sell local advertising to national advertisers granularized by metro market.
We have an agency using geo-data to do promotional campaigns for a Lasik Eye Center that has multiple locations. Whenever someone comes to their Web site, they’ll get information relevant to their location. They’ll be referred to the centers nearest them for contact and also only see advertising that complies with the state laws applicable in their geographic location.
Missing from this discussion/interview is the fact that people use the Internet for research and then overwhelmingly transact/fulfill offline.
Screenwerk 
December 18, 2008 at 7:05 pm
To clarify/reinforce, there are two examples cited above: 1) delivering targeted advertising and 2) serving relevant content. Quova, Digital Envoy, MaxMind and others are great for the latter, not for the former. IP geolocation can not offer accurate and consistent precision at a level below that of the metropolitan area (ie, DMA). This is due to a variety of technical issues which can be summed up in “remember when everybody on AOL dialup was located in Dulles, VA (but they weren’t?”
December 18, 2008 at 9:16 pm
IP Targeting is so January, 2008.
December 18, 2008 at 11:36 pm
totally
December 19, 2008 at 2:21 am
I’m not sure why I keep tracking things like this but I feel compelled to jump in. And sorry @ian, but you’re dead wrong.
I’m a co-founder of Digital Envoy and the guy who originally came up with the idea of IP geotargeting. IP targeting is in fact accurate down to the city level (not just DMA) and is that way world wide. At least, Digital Envoy’s technology is and given the fact that I designed it, I should know. Yes, you’re right that AOL users aren’t necessarily targeted correctly but they AREN’T targeted as Dulles, VA – at least not by Digital Envoy.
I would quote accuracy rates for Digital Envoy but given that I left the company in 2005 and they are now owned by someone else (the company was successfully acquired last year), I’ll refrain from specific details. I’ll just say it’s highly accurate to a city level on a world wide basis.
I guess when you create a technology like this, you always have this urge to talk about it and correct misconceptions. Even though at this point, I really don’t have a dog in the fight anymore.
Hope this helps.
December 19, 2008 at 6:28 am
Sanjay
I’ve been waiting for an IP father to come out of the woodwork as nobody has ever challenged me, so I’m glad you are. Please do explain, but note I’m not saying it’s a bad geolocating technique–it’s pretty much all we have to go on for a hard-wired PC–but consistency at the sub-DMA level is difficult to come by. DE might indeed be ahead of the curve, but I’ve never been able to understand how one can pierce proxy servers, load balancers and dynamic IPs with any regularity, and therein lies the challenge–you might know a user’s location down to a city (say) 50% of the time, but you don’t know which 50%, therefore advertisers might scratch their head. I’ve never seen any research substantiating claims about the granularity of IP geotargeting except for a self-audit from Quova.
To be clear, I’m not poo-pooing the technique by any means; hell, wish I had come up with it ;), but I do think that most interactive marketers do not understand how location is derived.
December 19, 2008 at 10:35 pm
Ian – heh, I’ve never been called an IP father before. Kinda funny.
So I can’t really go into the details of the technology for obvious reasons but you’re absolutely right. The mass of people have been taught that no, this isn’t possible. At DE we had this same issue when we were trying to sell. Basically we won customers over by showing that the proof was in the pudding.
A lot of the accuracy comes from having a good methodology and system to analyze the Internet. At DE we saw that because of IP address churn your accuracy could go down by 30-40% (at a country level even) if you were just 3-4 months out of date.
In terms of proxy servers, it really isn’t as big of a deal for the mass of users. There are obvious point examples (like AOL) but most companies/providers don’t do that. First because of reliability issues and second because of sheer cost. If you know the history of AOL you’ll understand why they have those proxys – it isn’t because of filtering, it’s because they built their original infrastructure on a non-TCP/IP protocol.
Also, load balancers aren’t an issue when it comes to identifying individual users. Users are never placed behind a load balancer – web sites/apps are. So that gets removed as an issue unless you’re targeting web sites/apps. If you are, it still isn’t an issue because servers are always geographically nearby when they’re pooled behind a load balancer (a load balancer never balances across vast geographic distances – Digital Envoy and Coyote Point Systems built a load balancer, Emissary, that could do this and even won an award for the joint product).
Finally, although dynamic IPs seem dynamic to us as end users, from a network perspective, they aren’t. And that’s what concerns companies like Digital Envoy – creating a map of the Internet from a network perspective and layering on geographic intelligence on top of it.
If you’re really interested in dry reading (and more details), you can check out the Digital Envoy patents. You can find links to them off my bio at:
http://www.sanjayparekh.com/about
Hope this shed some light on the subject. Clearly, I know way too much about this topic for it to be healthy. 😉