Forum Moderators: martinibuster
It was very frustrating and we eventually changed the field label to a synonym to stop the effect.
It has led me to this conclusion: AdSense's algorithm may be kludged to try to boost the appearance of "under-utilized" ads, that is, ads that are getting a very small number of impressions.
So a page that would normally get highly relevant AdSense ads, starts getting irrelevant ads because of even a single occurrence of a Kiss-Of-Death keyword, that activates a "fire sale" on the under-utilized Ads.
Just a theory, see how it works out for you.
thx
Think of it from the adwords point of view. You do not get first place placement just by bidding the highest amount as you do with Overture. Google looks at CTR also. They try to figure out what ads will make them the most money by taking multiple factors into account. I would guess that they also look at ad budgets, number of impressions available, etc. It could get quite complicated.
In this case they probably have some good sized budgets and/or high maximum bids that look more attractive to the algo than the other keywords on your site. Unfortunately in this case they don't have good targeting.
In this particular case AndroidTech, your web page resulted with that AdSense ad due to a fluke in the algorithm. While Google can deliver ads with good relevancy for most web pages, it will certainly fail on some other web pages. And this was an example.
I doubt that the AdSense algorithm can micro-manage the ad delivery on such a granular level. I think it just applies the same "imperfect" algorithm for every web page. If it can deliver relevancy for 75% of those pages, then that's doing pretty good.
If this is the case (and I'm speculating here) it might suggest that those 'fire-sale' keywords, if ineffective, will crop up less over time as AdSense keyword triggers...
What it also suggests is that if Google is conducting long-term split campaigns, it may be alternating between adverts that are relevant to site topic, front page topic, page topic, individual high-CPC on-page keywords, individual keywords which relate to infrequently displayed ads etc.
This might go some way to explaining why CPC and CTR rates seem to be absolutely haywire for everyone trying to micro-manage them... and more than that, why they seem beyond rational analysis for those webmasters who are constantly tweaking their page content in order to maximise CTR.
After all, if Google is trying out split campaigns and the page content keeps changing significantly, the split campaigns have to keep starting again from scratch.