Comscore’s Google Click Drop Prediction
Just a quick comment:
- Comscore is not always accurate, they are better than other sources, but not always right. They only observe a sample of Internet users, and their sampling error margins can be as high as 5% (they should start publishing that, similar to voting exit polls, etc.)
- Google clicks can go down, but revenue can still go up. For example, Google can decide to display less ads on their pages to improve the user experience by eliminating the low quality ads (e.g. by increasing the min-bid per keyword). This focuses the clicks more towards the higher CPC ads, and hence Google can make more money still.
I am not saying that Google will do bad this quarter, neither am I saying they will do well 
All I am saying is take Comscore’s click numbers with a grain of salt.
Cheers,
– amr
how do you get 5%? i think it is much higher error rate for the following reasons:
1. they underrepresent heavy net users who account for the majority of net traffic
2. their panels do not cover international users very well and a huge amount of the traffic on dotcom is from non-US ip addresses
3. no accounting for shared computers (kiosks, library)
4. virtually no coverage of mobile
5% error rate would be awesome if it were true - i would use them instead of weblogs if this were the case.
Comment by tom — February 29, 2008 @ 7:55 amAmr -
Glad to see that you are on your game and your post’s publication beat comScore’s own post on the subject:
Comment by Adam Durfee — February 29, 2008 @ 2:37 pmhttp://www.comscore.com/blog/2008/02/why_googles_surprising_paid_click_data_are_less_surprising.html