I’ve been meaning to point to these all week, but with the holiday, and all the catch up work after Web 2 and being sick as a dog last week, well. Anyway, this is kind of fun. The subject: Google’s search share (natch). The stats: well, Comscore says Google gained share, and Nielsen says it lost. I say, when can we get a stat provider we can trust? SEW does a nice job of wrapping it all up here.
Meanwhile, Hitwise (whose data feels most suspect of the three given the results Danny tabulated in the pic to the left) has a nifty chart where you can chart Yahoo v Google for overall audience here.
In the case of MSN – many users are just using it because it is default on Internet Explorer….The same for AOL.
But in the cases of Google and ASK many users have to actually TYPE their addresses in the browser.
Yahoo probably receives a combination of users typing it in the address – and users from the default email box in Yahoo Mail, News and other services.
Butm one has to admire the way Yahoo is constantly attempting to remain competative via their portal and web 2.0 freebies.
It would be interesting to discover how may users actually, initially put Yahoo.com in the browser address when they want to search – and why they choose Yahoo Search over Google search?
I’m not sure how these companies get the data they are publishing, but it seems to me that you will always get biased results to these kinds of numbers if you require users to install or download something. Taking data from referals or server logs is equally biased, Unless of course it came from the search engines themselves. It would be interesting if someone like Ask or AOL (I mention them as they may not feel they have as much to lose) were to publish actual numbers…
I don’t understand why these search engine representatives do not do a press release about this information.
I believe that Hitwise uses ISP-based (anonymous) traffic logs. It appears as though AOL and “other” networks are under represented in the Hitwise sample, which could account for some portion of the higher Google %.
Hi Lance,
That is interesting to hear how Hitwise works. Once again I see a huge opportunity for bias. If they include logs from AOL’s isp network we would probably see a bias towards AOL search sites, likewise if they don’t include AOL you would also get a bias as the number of users on that network is (i’m guessing) significantly large that it’s numbers could have an effect on the final result. Basically the problem is getting representative sample of the data, So far I haven’t been able to think of anyway to get one. Other than (as in my earlier comment) search engines being willing to publish the ‘real’ stats.
In answer to your question Dave, I suspect they don’t want to risk looking smaller than the competition, just like google has stopped publishing the number of sites in their index.
That also had problems with how you count “sites” do you include spam? etc. I imagine if they were to publish a “number of searches” this week stats you would probably hit issues of what a search is as well. Eg. Do you include people searching for the same thing twice? What about API requests, special types of searches, Yahoo Shopping searches, Gmail searches?
In otherwords, It would probably be too much of a bother getting everyone to provide equivalent data anyway.
– James
Hitwise numbers may be different from ComScore and NetRatings but Hitwise matches the referrer data I’ve historically seen for most websites. This could be due to the individual sites that I’m familiar with indexing in Google better than other search engines but I doubt that is the case.