How Does Local Search Work?
Local search is based on several different methodologies, including the science known as geolocation, the science of trying to figure out where the heck a computer is, geographically speaking. How does the search engine figure out whether a computer is in Colorado or Florida?
Well, local search generally works in a few basic ways. Different services use different combinations of these methods.
If someone types dentist new york, the search engine can be pretty sure that the person is looking for a dentist in New York, not a dentist in Oklahoma City. Simple, eh?
Partner and localized sites
Search services can guess at a location based on the website someone is using. If someone is searching at Google France, it’s a good bet that the person is in France; if someone searches at Yahoo! UK, that person is probably in the United Kingdom. In other cases, partner sites can be even more specific, and related to a particular region or even city.
Internet Protocol (IP) numbers identify computers on the Internet. Every computer connected to the Internet at any moment has a unique IP number.
With information being sent to and fro — from your computer to websites and back — there has to be a way for the information to be "addressed" so that various servers on the Internet know where to deliver the information. Thus, every computer connected to the Internet has an IP number, or IP address.
In some cases, two or more computers share an IP number, as in a situation in which a house or an apartment is using a cable or digital subscriber line (DSL) connection, with several computers connected through a single router; but for the purposes of figuring out location, this information isn’t important, of course.
In some cases, computers "own" a particular IP number: Turn the computer off now and turn it on next week, and it has the same number. This is known as a static IP number. Often, however, computers share IP numbers. Log out of a dialup account now and in five minutes dial back in, and your computer is assigned a different IP number (known as a dynamic IP number).
That IP number is shared among many computers, but only one computer can use the number at any particular time.
Take a look at this IP number: 188.8.131.52. This number uniquely identifies a particular computer in Colorado. If a server sends a page to that address, the page can go to only one place because at that moment only one computer on the Internet is using that number to identify itself.
It’s like a telephone number. Every telephone in the entire world has a unique number (when you include the country code). Pick up the phone and dial the full number, and there’s only one telephone in the world that you can possibly be connected to.
An IP number is a hierarchical system. A block of numbers is assigned to a particular organization or company; that organization or company then assigns blocks of numbers to other organizations or companies, which can then assign their numbers to different organizations or companies, or divisions within a company, and so on.
Consider, again, 184.108.40.206. This number is "owned" by Comcast Cable Communications, a large American cable-TV company. In fact, Comcast has a large block of numbers: 220.127.116.11–18.104.22.168. Within that large block lies another block that Comcast uses in Colorado: 22.214.171.124–126.96.36.199. Clearly, 188.8.131.52 lies within this block.
Local-search marketing with IP numbers isn’t perfect; it’s definitely an imprecise science, for a few reasons:
You can’t assume that a number assigned to a company in a particular area is being used in that area. It’s possible for two computers using two IP numbers that are just one digit apart — 184.108.40.206 and 220.127.116.11, for instance — to be thousands of miles apart. An IP number block assigned to an organization in one area can be used by many different computers in many different locations.
Dynamic IP numbers are "here today, there tomorrow." When you dial into a network and are assigned an IP number, you could be in California while the computer that assigned the number is in Virginia. When you log off and someone logs on and takes your number, the new computer might be in Wyoming.
Still, geolocation is getting better all the time. Although the authorities that assign blocks of IP numbers provide very basic geographic information, this information can then be combined with other clues, such as hostnames.
For example, it’s possible to trace the path taken by a communication from one computer to another and get the IP numbers and hostnames of the servers between the start and destination computers. Hostnames sometimes contain geographic names that help IP-location systems figure out where they are.
Another clue: Some major Internet service providers (ISPs) assign blocks of IP numbers geographically, so when you crack the code, you can figure out where people using that ISP actually live. Using clues such as these, geolocation engineers at various companies specializing in this science can fairly accurately locate IP numbers.
So the system isn’t perfect, but it’s close much of the time. In any case, it doesn’t usually need to be spot on. If you’re searching for pizza in your town, for instance, Google doesn’t need to know your exact address; it shows a map with pizza places, and you can zoom in to your specific location if you want.
Google lets searchers actually specify where they are. Click the cog icon at the top right of the search page and select Search Settings. Click the Locations link in the left column, and you’ll see a box into which you can enter your street address. The only problem is that the vast majority of users don’t know about this setting.