Are Google Results Affected By IP Address or Web History?
I have heard many times that Google delivers personalized results based on both your IP address and your web history. While this may be true, it may not be statistically significant for a clients’ highest volume organic keywords.
Let me share my findings from an hour or two of research today. My process was to perform manual search queries for a set of 10 high priority and high traffic volume keywords related to new and used automotive searches looking for a particular website to compare the rankings. To test the theory of IP addresses delivering different results, I used a proxy server with a Denver, Dallas, and California IP Address and verified that Google would indeed identify the search as being from that particular city. I chose Denver because that is where I am currently located, Dallas because it is where I host a virtual server, and California because it is where the client is located.
Before I started, I downloaded a Firefox plug-in from Yoast that automatically added the correct query string variables to remove personalization from my search results. I was ready to get started!
Using Denver as a control group, I ran the 10 keywords (keywords ranged from ‘cars’ to ‘new cars for sale’) and I received an average ranking of 9.36. The same keyword searches in Dallas received a 9.09 and California showed a 8.82. This didn’t really surprise me as the client and the website is located in California and Google may give preferential treatment to the rankings in California.
What did surprise me is that Denver, Dallas, and California had no difference in ranking on 4 of the 10 keywords. For the 4 keywords with ranking difference, California showed small improvements on 4 of the keywords and actually returned worse rankings on two keywords. There was actually very little difference between Denver and Dallas which could be due to their proximity.
Next, I wanted to test the effect of Personalization due to my web history. I logged into my web history and checked and found around 100 searches related to cars and over 15 searches directly related to the client website. To help the personalization, I did a few more searches, clicked around the site and made sure that the Google Toolbar would feel like I really liked the site. With previous SEO research being done on the site, the number of previous visits to the site should be very high.
Across all 3 cities, my Non-Personalized search queries had an average ranking of 9.27. I logged into my Google account and repeated the searches via the various proxy servers and calculated an average ranking of 9.09. The lowest calculated average ranking score was posted through the California proxy server while I was logged in at 8.82.
One of my thoughts during the process is that long-tail searches such as ‘los angeles used car search’ might have been a better control. But when I tested a few of these keywords, I didn’t get any more difference in the rankings. I do think that Google does show different results for different type of Geo-Modified searches based on where you are located, but that isn’t what I was testing. I was checking to see if a particularly strong site would be ranked differently in different geo-locations even though Google may actually change the information around the core set of rankings. While not tested in this research, it is an interesting thought exercise.
SO…basically, all of this research shows that rankings can be affected by IP Address Geo-location and Personalization. However, the statistical difference on this test group isn’t strong enough that it appears that a normal Google searcher would experience significantly different search results across different locations.
Comments are closed.