Forum Moderators: open
For the worst affected keywords I took a sample of 425 keyphrases where the filter has removed between 95 and 83 of the top 100 pages in the SERPS. These statistics are available on a well known site that allows you to test the effect of the filter. (I ignored keyphrases where more than 95 out of the top 100 pages have been removed because I thought there might be more than one filter coming into play for them.)
For the least affected keywords I took a sample of 235 keyphrases where only between 0 and 3 pages have been removed from the top 100 pages in the results.
I put these keywords through the Adwords traffic estimator and looked at the estimated traffic, average cost per click for top Adwords position and estimated cost per day for that traffic (for one ad).
I removed keyphrases that were generating no measurable traffic as that would have included spurious results in the least affected keyphrases.
The results are (costs in British Pounds):
Worst affected keywords:
Least affected keywords:
i.e. On average the keywords in this sample that are worst affected are likely to generate 6 or 7 times as as much Adwords revenue as those least affected. (Or more, as there are likely to be more advertisers for those keywords)
Conclusion:
It is too early to leap to conclusions.
Demonstrating a correlation is not the same as demonstrating cause and effect. Google may be selecting the keywords for the filter by some entirely different method that just happens to have a correlation with Adwords revenue.
We must not muddle cause and effect. The higher Adwords traffic and bidding for the worst affected keywords may be an effect of the filter and not the cause, as users can only find what they want in Adwords, and advertisers may be bidding more because they can't get traffic from the SERPS. If this is the case, then the increased Adwords revenue may be an unintended side effect, but Google is still laying itself open to justified criticism.
My samples may be flawed.
The filter may be triggered by part of the keyphrase, but Adwords estimates are for the whole keyphrase, so I might not be comparing like with like.
Help
We need more statistics, but I do not have the time. Has anyone else done any analysis? Can anyone else analyse a larger sample?
Has anyone measured user (searcher) satisfaction with the new Google SERPS? Has anyone been out and interviewed 100 people in the street? Have you polled you colleagues at work? There must be lots of members with colleagues who know nothing about SEO or even that Google has changed - Have they noticed any difference in their searches?
