Getting different positions on Google.com, Google.ca or Google.co.uk is generally well understood in the industry and always taken into account when measuring the performance of a site in another country.
What is less understood is the difference between positions whether the search is coming from Canada or another country (whatever Google version is used). We noticed a while ago that there could be a huge difference in positions between searching for « keyword123 » on Google.com in Montreal and searching for « keyword123 » on Google.com in New York.
After making sure it wasn’t discrepancies between data centers, I combined the following data using Google Ad Preview Tool:
Based on that data, here’s my conclusions:
- It becomes clear that the most determinant factor is the heavy usage of geo keywords (ex: quebec, canada, etc…). Site # 4 use a little bit of those keywords but none on its home page title.
- When geo keywords are used, the server location seems to have little influence
- What? No difference between Google.ca & Google.com? No sir.
Based on other tests I performed, here are more conclusions:
- Some said that long tail keywords where less affected by the differences in rankings. I tested a lot of keywords and I got mixed results. I personally think it has nothing to do with the number of keywords but rather with the precision of the search: if your search if broader and less precise, you have more chance of getting bumped by a greater number of competitors.
- If there’s a geo related word in the search query, the rankings are the same in Canada and US (ex: computers Canada yielded very similar SERPs in Canada & US)
So what does it all means? It means that google.com applies a geo filters on its results depending on the IP where the search is originating and that in Canada, we just can’t rely on Google.com data to evaluate the rankings of a site in the US.
The poor performance of sites using geo related keyword is also really problematic for tourism sites that want to promote specific activities in the United States through organic search listings.
A couple of facts make me think that this behavior is either new or has never really been discussed. First, this thread on webmasterworld.com is only a couple of months old (and has heated up only a couple days ago). Second, I talked to some of my friends in the business and neither of them seemed too concerned about the matter. And finally, the differences in ranking seem to be only with Google.com and not with other TLDs of Google (and we know Google.com usually gets new stuff first).
To measure the real positions of your site, I would either suggest using the ad preview tool of Gooogle or more realistically, as Seomoz.org recommended (you need to be a member) using a proxy with your ranking tool.
