It is generally accepted that “clicks” are biased by a terms ranking position, and snippet (ie its boldness and repetition). The user of course is biased not the domain or the click!
Microsoft carried out a study and then a couple of carefully controlled (and fully documented) experiments to prove their hypothesis:
The study basically looked at user behaviour vis a vis search engines at an overall level. They did a series of tests over the same time period in 2009 and 2010 on a particular search engine to see if changes in the displayed results eg changes in ranking of domains were reflected in changes in click-throughs. They weren’t and were approximately the same by domain name which indicates a possible bias in favour of specific domain names. They also noted a concentration on fewer domain names and corresponding click-throughs in spite of the continuing increase in the overall number of domains on the web. They also confirmed this bias applied in tests involving two markedly differing search engines.
The first experiment they did was, in their terms, a Pepsi / Coke type test. ie differential labelling (in this case a domain name) attracts a preference or bias irrespective of content. They reckon they did prove that the perceived relevance of search results can be swayed by domain names.
The second experiment was a rather complex one involving “currency” term searches that confirmed that users do exhibit a systematic and consistent bias over domains.
As a result of their work Microsoft suggest that search engines click models, which generally focus on the three factors that influence clicking ie the users query, the position and relevance of the results, need to be revisited and refined.
The way forward Microsoft suggest is some form of domain bias calibration for incorporation in search algorithms. We think its almost a brand calibration.
If unchecked the likely outcome of this increasing trend Microsoft discovered is that it leads web users to concentrate their visits on fewer & fewer domains which debatably is not conducive to the health & growth of the web in the long run.
Our view is that this bias includes not only the domain name but also the toplevel domain suffixes (TLD’s) at both the generic (g) and the country code (cc) levels. Hence there is user bias towards a .com versus a .org and in the UK bias in favour of .co.uk versus.org.uk or in some cases over a similarly named ,com or ,org.
The country codes used widely for generic purposes like .co and .tv will generally start with a negative bias versus a .com or a .co.uk although in certain very limited cases a reputation establishment can override this negativity eg bit.ly and t.co
Also we conclude that there is definitely a huge bias towards shorter and hence more memorable domain names!