Personalized search favour popular sites?

Google has, as many know, introduced personalized search, where the search results are based on the individual user’s web-history with Google.

Personalized search can be switched off, but I think it’s fair to assume that a large proportion of the Google users will be using personalized search, thus not seeing the basic SERPs without personalization.

Implications for SEO

Google seems to maintain the basic ranking algorithm, and the personalization based on the web history, is used to correct the rankings when the user make the search.

We will therefore see more variation in the rankings accross the different users, thus making ranking checks more unreliable than before. This is however not a new problem, as rankings also are personalized for all the users who are logged in to Google, while they perform the search.

The interesting point is whether the average personalized results are biased compared to the non personalized result – i.e. does personalized search favour certain sites.

Ranking Bias

If the personalization is based on the non-personalized result and the web history, there would in my opinion be a bias in the average rankings, as the more popular sites will appear more often in the web history, thus improving rankings for these sites in the average personalized rankings.

This could lead to quite a large ranking bias, where the popular sites with a lot of traffic on average will move up in the rankings, and the small sites with limited traffic will lose ranking.

If this is correct then popular sites on average will rank better than the in non-personalized SERPs, and the small sites will on average rank lower.

Are this bias really important?

If the ranking bias is significant then Google will favour the large and popular sites with the introduction of personalized search, then it will be increasingly difficult for new and small sites to break trough.

These sites will struggle to get a top ranking in the personalized results, as the web history will create inertia in the rankings, that will work against changes in the rankings. The user will click on the usual sites thus improving or maintaining the rankings of these sites.

How big are the ranking bias

The experience with personalized search is so far limited, and the ranking effects of the personalization seem to be quite limited in most cases. But more tests and research are needed to make firm conclusions.

Google log-in personalization has been online much longer, and in my experience the personalization effect can be quite large with the Google log-in. I can’t really see why Google in the long run would like the put less weight on the web-history in the general personalization – and the personalization effects could therefore be quite big also without Google log-in.