As Bing and Wolfram Alpha have shown over the last couple of months, big sums are still being gambled on finding a chink in Google’s armour. Of course, search engines can only maintain their market share by serving the most relevant results. Historically this meant matching results to a query through algorithmic calculations. Increasingly though, the search engines are trying to provide better data to their users through other things – such as timeliness and geography.

How it works
Geotargeting takes place in various ways. The most obvious is by tailoring search results by requests made by the searcher. Someone searching for ‘underpants’ will be served a very different set of results to someone searching for ‘underpants London.’ In the first case, the person might just be looking for a definition of what underpants are (perhaps they’ve grown up in a very liberal environment). In the second, they are more likely to be searching for shops to buy them.

Each of the major engines now has its own mapping software – populated with local results through advertising channels. Google have taken this to its logical conclusion by embedding map results in its main results wherever there’s a geographic element to a search.

Secondly, they can get an idea of a person’s location by information that has been supplied to them by the user. Google and Yahoo, for example have accounts which are used by people who sign up to their ancillary services like email. Often, users will even supply their postcode as part of the sign-up process, allowing the engines to give highly tailored results. A common example for Google account holders for example is to see local results on a map within the main results whilst logged in – even if they haven’t searched using a place name. When I search for a film while logged in with my Google account, I’m shown local film times. Neat!

Thirdly, the search engines index content according to the language used. Google has country-specific indexes of content in French, Chinese, Arabic etc, which are served to users by datacentres based in the relevant country. These use a variety of factors to determine which content to serve to users – from the geographic location of the servers to the measures of relevancy that match keywords in other languages in the same way they match English. Obvious clues such as pages in specific languages and the location of the person in whose name the domain is registered are factored in to these results.

Finally, they can use context supplied by technology. IP addresses and traceroutes can, to some degree, provide clues about where the user is located. You’ve probably come across results on Google where a map is embedded in the results even though you haven’t specified a location. You’ve also probably had the same experience as me: I’ve searched for sweet shops and for some reason known only to itself, Google chucks in results near Liverpool or Nottingham or Turkmenistan. It might give you a clue as to how certain Google is about this feature that these “guessed” results typically show lower down the page than they do for “certain” matches. That hints at how wildly variable this kind of targeting can be.

Working with Geotargetting

Much of geotargetting is out of your control. If your user’s IP address places them in Liverpool so far as Google is concerned that’s pretty much hard cheddar if both they and your business are actually in Hull: if Google wills it, so ye shall see results for Liverpool. And if someone is signed in with their Yahoo/Gmail account and looks for something then similarly you’re plumb out of luck if they’ve entered 90210 as their postcode.

Where you can score is to make sure that your business is properly listed with the search engines’ mapping services. Through Google AdWords, for example, you can combine keywords with a map listing to give yourself a great shot at appearing top of the map listing for a keyword search.

Serving Results for Foreign Language Queries

A common tack taken by site owners is to use IP addresses to serve different content to visitors – typically pinging them off to a French language version of the site because that’s where IP address looks like it’s located. There’s an obvious danger in doing that because of the inaccuracies inherent in IP location tracking. And if you’re a big English language site, do you really want to serve your customers your website in Greek just because they’ve taken their laptop on holiday?

Secondly, you run the risk of tripping an algorithmic penalty because you’re basically cloaking content. While your motives may be good, and the results good for a lot of users, it’s a chance you probably don’t want to take. Losing a few days or weeks worth of traffic while Google try to sort out whether you’re UA or IP cloaking and what your motives are could be pretty costly.

Putting flags or text links (hint: keyword links in your target language) on your site to indicate other languages is a safer option.


Target local languages with localised text, and make it accessible with obvious links

If you’re after something more esoteric, like ranking for a term in a foreign iteration of the search engine, there are a shedload of things to shovel into the boiler: ideally, a full foreign language version of your site on a country-specific TLD and hosted in the target territory. If that’s not practical – and very often it isn’t – you can rank by having language-specific content and building keyword links in the target language from sites in the relevant territory. The alogoritms are sophisticated enough to make judgements about what content to serve.

"Of course, leaving your link request emails to the vagaries of Babelfish might not be the best idea but other than that, it’s simples."