I've raised it as an issue with Google and spoken to them several times over the last few days. They seem (genuinely) to be as baffled as I am and have given their usual answer that its been escalated to one of the technical specialists and that they'll come back to me.
Having also now checked other adgroups in different campaigns I can see that its not an isolated issue and the problem exists on random keywords throughout my account (but always high volume / high competition). If I was at all paranoid I would suspect that it may be tied up with whole quality score cull and that Google is secretly inactivating keywords in competitive sectors (without flagging this to advertisers) based on other companies with bigger wallets being prepared to pay more. If this were true (and I'm not suggesting it is) it's almost like being penalised for being successful at doing what Google have always preached (i.e optimisation) - the keyword is highly relevant to the ad and the landing page and thus has a high CTR and low cpc - it also passes the quality score test but now they decide that they'd like to eliminate you from the space altogether on the basis that others are prepared to pay more.
Could this be the next step in filtering out affiliate sites (as this one is) from adwords? 'If our quality score algo didn't get you then then we'll just simply turn off your keywords and not tell you!' A scary thought but surely can't be true - can it?
There must be a logical / technical reason for this and wondered if anyone else had experienecd similar and if so what the explanation was and was it resolved.
Many Thanks
Ross
My assumption has been for a long time that Google measures the profitability of the entire search results page for each keyword users request. And that the ads are rotated "experimentally" on the results page to see if they can boost profitability by reorganizing the ad placements on a page. This experimentaly reorganizing gives ads in a lower position an opportunity to replace the premium position ads. This experimental reorganization could also include removing an ad or ads from a page completely.
This sort of rotation might be the only real way to control for the selfufilling higher CTR's that come with premium positions. For example a long term bidder on a keyword may have gained the number one position a long time ago and achieved a CTR of 28% with a bid of $.25. A new bidder may have a newer ad in the third position with a CTR of 8% and a bid of $.25. Together the two keywords achieve a CTR of 36%. The 8% CTR in position 3, may be a good predictor (don't forget this is all speculation on my part) that if it was in position 1, it would get a 31% CTR. The only way to know for sure would be to put it in position 1. In continuing the experiment, lets say initially the #1 and #3 keywords are swapped in position. Then lets say the new bidder's keyword achieves a 39% CTR in the new position and the old keyword achieves only 3% in its new position. Now Google is getting a total CTR of 42% on the 2 keywords combined both bidding $.25, which increases profitability. Another possible conclusion of the experiment is that the algorithm now "knows" that the primary reason the keyword was achieving the ctr of 28% was the selfulfilling benefit of holding that spot. The experiment may continue with the old ad being pushed further down the page, each time yielding an improvement in Google's profitability for the page.
To continue on with the experiment the ad maybe "removed" from appearing at all on the page to see how profitability is affected. I am not sure how they "remove" the ad, but during this phase of the testing, the ads diagnostic tool may treat the ad as if it doesn't exist, which explains the other diagnostic explanation.
Could be more to it, but it explains why impressions, etc might have dropped, and in some niches, why some things seem to have been hit, but others not, depending on keyword.
However - I have noticed other weird behavior like you are explaining lately. For example I'll do a search for a certain keyword, and only 1 or 2 ads will show up. Then when I hit the search button again, an entire page or more of ads will then show up. And it will fluctuate back and forth every time I hit the search button. I think they are experimenting with ad position, and broad matching keywords - which to me seems unethical to make us pay for their experiments.
Just to clarify, these keywords are all exact match and in the broadband sector (and they're definitely 'sales' orientated rather than information based)
I will feedback on Google's response (still waiting)
btw - we've operated in the broadband sector for some time and its possible to buy keywords a lot cheaper than £2.50.
One company pays 30 USD and gets page 3, the other pays 5 USD and gets page 1 number 1. I've seen some huge accounts and they all tell different stories. The dynamics of the system really are quite amazing.
I've always wanted to run an anonymous price index, but I doubt advertisers would play ball!
The min bid isnt kicking in and my Google response when I asked about this was along the lines of roughly that for my keywords and ad text, terms I show for will match my vertical - sounds like a similar method used on the content network, so they are potentially classing me out of searches in a we know better manner without the previous you must pay stupid money. Im pretty suprised they didnt do that the first time around after the QS update as its harder to detect and less abrubt spin doctor wise.
The official reasoning being that not all variants of a keyword are applicable to your site. I think the term is 'slow death'?
