How Does Google Know Where You Are?
It is seven years ago this month that Chris Sherman first asked me to speak on a panel in London on the subject of multilingual or international search – with an emphasis on the SEO! I’m delighted once more to be presenting on “speaking in tongues” and even more pleased to be following SMX in […]
It is seven years ago this month that Chris Sherman first asked me to speak on a panel in London on the subject of multilingual or international search – with an emphasis on the SEO! I’m delighted once more to be presenting on “speaking in tongues” and even more pleased to be following SMX in London with an International Search Summit where we can dive deep into the the whole subject.
As always, I wanted to bring something new to the party and it occurred to me that we talk a great deal about geo-targeting our own sites, for instance on whether you should use local domains, sub-domains or folders and does local hosting matter — in fact this was the very first question I was asked in London all those years ago.
However, we don’t so often dig into how Google does it for the results themselves.
How Do The Filters Work?
Let me clarify what I mean. When you put a keyword into a Google search box, how does Google filter its results for you based on the country you’re searching from and the language you’re using?
If I polled a bunch of experienced SEOs in a room, bearing in mind that it would take at least four to change a lightbulb, I’d like to bet that the majority would say your IP address would be the most significant factor — but is it in fact?
With my team, I carried out some testing and discovered that it seems the IP address is not actually the most important factor.
Firstly, we need to understand what Google is trying to achieve. For every query, Google wants to try and deliver the most relevant result and is looking for every possible signal to help it predict what that result would be. The location of a user is a very significant clue to the purpose which is in the searcher’s mind.
Where You Search Plays A Key Role!
Take for example a search for a relatively vague term such as “Lufthansa”. If I’m on the USA’s west coast for instance, there’s a good chance I’m looking for a flight. If I’m in New York, Berlin or London I might be looking for a flight or financial news about the company.
If I’m in the USA or UK, I probably want my results presented in English, but from Germany, I’m more likely to want to see them in German. (Incidentally, I have no connection with Lufthansa other than flying with them from time to time.)
It also has to be said that as of today, I don’t have a definitive answer to all of the possible questions — I’d be delighted to receive thoughts and contributions from readers. But this is how I think it happens.
Domains Win Over IP Addresses?
Firstly, Google wants to define what country you’re in. Getting that wrong could be seriously detrimental to the company’s performance globally — so this has to be top of the list of priorities.
Fortunately, IP addresses on a country level are generally relatively accurate. Nonetheless, Google seems to rely more on the typed-in domain name than the IP address.
In the above graphic, we’re looking at the movements and differences which take place between the Google.com domain and that for Google.de. As you can see, there are some fluctuations which represent the typical shifts based on testing which took account of results coming from different datacenters.
Interestingly, the IP address for both sets of results is from the UK. We’ve deliberately used a keyword which is specifically German so the language effect is not playing too much of a part.
The above graphic is further proof of the pre-eminence of the domain. Circled are three Austrian websites with .at domain names. The search, however, was done from the same UK IP address used to compare the US and Germany above.
You might say, you’re confusing this by using a UK IP address — but in fact this is not the case. When a German IP address was used to compare German versus Austrian results the differences were in fact the same. The selected domain name is clearly a key factor.
Google does, of course, use IP addresses for more hyper-local results (San Francisco versus New York, for instance) but even for this purpose, they must clearly be using additional data as the accuracy achieved in the UK exceeds the potential of just IP addresses alone. They also redirect users to the relevant Google domain so to this extent IP addresses are important.
The above graphic shows a comparison for a search for “Lufthansa” carried out on Google.co.uk and Google.de where the results are utterly different. This didn’t change when moving IP addresses about either. What’s interesting here is that the pages presented in Google.co.uk are English-language pages from Lufthansa and from Google.de are in German.
Note that these are from the same IP and PC — so any language settings are being overriden by the knowledge that “Lufthansa” can sometimes be a German term and sometimes and English one.
As the above illustrates, it’s our contention that you have to treat keywords as having a kind of invisible “language tag” which determines which results are presented. Our example is the keyword “casseroles” which happens to have a different meaning in French to English. The images displayed from Google images for Google.co.uk show meals, whereas the French images show the casseroles saucepan. So, Google domain and keyword language tag are the key components which determine how the web is filtered.
By the way, if you’re still wondering why it takes four SEOs to change a light bulb, you have to bear in mind you’d need one to check the compatiblity of the component, one to handle the re-directs for the new light juice (turn the switch on and off), one to hold the door closed for safety (don’t want Google’s crawlers to arrive at the wrong moment) and a web analyst to spot if it lights up!