Google Goes Las Vegas from Robert X. Cringely at PBS looks at the anecdote of someone who was paying $0.10 per ad to promote his site and then tried to test new ideas with a different site, coming away perplexed about why paying more for the second site generated LESS traffic. Are things rigged, Cringley asks?
The person he writes about decided to try testing ideas on getting better results with an entirely new web site. When he created ads for this site -- using the same terms as the old site -- he was told he had to pay $1.00 per click.
It's not clear whether he actually had to pay that full amount or if this is what Google was recommending he pay that amount (what you pay and what Google suggests and actually bills can be different). But he did pay that amount and got more clicks than his main site.
Then he dropped to $0.40 per click, and his clickrate plunged. Why didn't the test site still do better than the main site? After all, it was paying four times more.
Cringely says he has "no idea" what this happened, leading to his main point. Since the system has a lot of things to calculate about how well an ad does, more transparency would help those who might assume it's just rigged to benefit Google.
Of course, the system's always been rigged to benefit Google. Ranking ads on the old formula of CPCxCTR did help promote relevancy, but first and foremost in my mind, it also ensured Google was giving the most play to ads that were making it money.
Recent changes -- as covered in New Google AdWords Bidding System Live For All and Goodman Revisits AdWords Changes & Importance Of Clickthrough -- introduced the quality score factor, but exactly what's in that isn't certain. That's concerned some advertisers. Certainly a clearer explanation would help. Cringely's article points this out. John Battelle calls for the same here, and Andrew Goodman, of course, has been wanting that for weeks.
For all we know, part of the quality score might be rewarding an advertiser with a longer history of performance with Google. Since this advertiser set up an entirely new account for an entirely new web site, the quality score might have counted against him in that instance, since that account had no real history.
It would also be interesting to know more about exactly what wording was used in both ads. Is the "main" site a more "trusted" URL to users? That has an impact. More transparency from Google would help, but knowing more specifics about this advertiser's situation would also be useful.