Geo-Location Basics

Every connection online has an IP address (though sometimes individual machines may be sharing the same IP Address due to network setup), which acts as a unique identifier for that connection. This allows routing of data packets across the local/global network. Think of it like your house address, but for your computer online. This is abstracted upon with DNS, allowing us to reference certain IP Addresses of servers (like webservers) by a name rather than numbers. An example would be Google.com, which is a DNS alias for 74.125.225.81 (as of the time of writing). The address 74.125.225.81 belongs to the server that is hosting the google.com website.

The most common methodology of geo-location look-up is IP Lookup. This is where a system takes an IP address and checks it against a matrix of IP/Location data that is maintained by several providers (1). This commonly consists of: Continent, Country, Region/State, DMA, PostalCode/ZipCode, Latitude and Longitude. The broader the area you are trying to determine, the more accurate. For example looking up an IP’s COUNTRY would be more accurate than the CITY an IP belongs to. The same continues down, so City is more accurate than Postal/Zip.

A term that is often heard when talking about Geo-location is Reverse IP Lookup. This is really a miswording, as what is really meant is Reverse DNS Lookup. This is when you lookup the Domain Name associated with an IP. So if we did a reverse lookup of 74.125.225.81, we would get google.com, except Google has not set a reverse DNS entry in their DNS Zone setup (its not really needed unless your sending email from an server with that IP). Reverse DNS lookup is often used to gather additional information about the user location, especially in the case of Cable/DSL lines, where there will often be a value that gives location information such as City/State/Network Switch in it. An example would be Liquidus’s connection (66.166.70.42) shows up as h-66-166-70-42.chcgilgm.static.covad.net. You can see by this string that “chcgilgm” refers to Chicago Il. Of course this requires parsing a string for specific formatting, that could be different between carriers, but most carriers at least wont change it often due to impact on internal process/documentation.

Another additional level of depth of location information can be gathered through the AS number that is assigned, which can be seen with Reverse DNS lookup. This number represents the business entity that “owns” (owns is in quotes because often it is leased) the IP address. When the information in this packet is correct, this is one of the most accurate location signifiers. It can be as specific as “Science Building at Carnegie Melon”. However, the issue is “when its correct”. Often the information is kept at a higher level of the business structure (say Corporate HQ) to cut down on the maintenance cost of constantly updating micro-level information when these IP addresses change hands. There are some 400,000+ AS numbers available today, with a potential to grow to 4 Billion.

Overall there are bunch of different methodologies for gathering a user’s geo-location, and very few of them are regulated thoroughly, and so the accuracy level – at least with IP-based – suffers due to varying business-level models/maintenance. Remember that with any geo-targeted/messaged campaign, you should always allow the user to correct the assumed location. In our Bannerlink product, we allow the user to enter/change the zip code to ensure the most accurate information is displayed.