Hi I just wanted to moan. I'm in data collection, and as many of you know, we've been hit by Google for some time now. Anyway, since being slapped with ridiculously high minimum bids last autumn, we thought we were going to have to abandon AdWords. I turned on the Conversion Optimizer and guess what?! - ALL OUR KEYWORDS INSTANTLY GOT 'GREAT' Quality Scores. I mean ALL - every single one. So our registrations poured in and things have been great since November. Yesterday I logged on and suddenly the entire Quality Score column dissapeared from all my adgroups and our minimum bids were all over the place. Nasty. 75% drop in traffic. Bad. We have a free dating site, but the keyword freedating is now given a poor QS. Just to show the idiocy of the algorithm, the keyword jewish dating (which does not appear on our landing page) has a POOR QS and a minimum bid of £2.50 while the keyword romanian dating (which is also NOT on our landing page or whole site) has an OK QS with a minimum bid of £0.10. In my experience the Google QS algorithm is a complete joke. Thanks for listening.
The QS is not solely based on the presence of keyword on a landing page. It's based on a number of things - including your keyword performance in relation to the historical performance of that keyword across all advertisers. Understand what you're knocking before you knock it
Actually the historical performance of the keyword across advertisers is a very minor though salient factor, as was told to me today by a Google rep. Only an idiot would make that a dominant factor in an algorithm, as it would eliminate given keyword's chances of ever performing well on newly created landing pages. Remember the football adgroup that had the keyword 'hard core porn' in it, which had a QS much better than plenty of football related keywords? The Google rep said that this must be a bug. Thanks for the input.
Hey, try Google's keyword tool (adwords.google.com/select/KeywordToolExternal), and enter the URL of your landing page. They'll extract Relevant keywords from your site and list them nicely for you... Seriously, there's no need to worry about keyword quality scores because Google themselves are practically telling you which keywords are relevant to your site. Think about that for a second...
it is all about the money to them, and not the QS, and as a result of that simple fact we all get the BS, plain and simple. A few years ago you could bid on the keyword cock for 1 cent and now it is $10.00. come on now..... maybe all keyworkds should be the same price to level the playing field no matter what the category.
Don't believe everything you're told...and that answer in relation to the QS may be partially true, but needs to be put into the proper context. It may have less of an impact in high volume situations than it does in medium to low volume situations where QS updates are less frequent. So in your situation maybe it is "very minor" but that very minor element just pushed you to poor. Lastly, "very minor" is relative. I wouldn't put a very minor headache and a very minor heart attack on the same level. Only an idiot would use all that historical data to determine relevance??? If you understand how that data could/would be used you would understand that your statement "it would eliminate given keyword's chances of ever performing well on newly created landing pages" is completely false. Anyway, my point was that you were taking a somewhat complex subject with a lot of potential variables and apparently misinterpreting what you were seeing and arriving at false conclusions. I was simply pointing out that the quality score is more involved than the inclusion or exclusion of a keyword on a page or site. No I don't.
Thanks for the suggestion but I've used this tool many times. Quite often it too suggest irrelevant keywords.
Thanks for the reply. "Only an idiot would use all that historical data to determine relevance?" Yes, if relevance is really the goal. A word may exist for a long time only to change meaning or usage at a later date. This change could well be very niche. It seems rather obvious to me that within most distributions you'll have some data points that are far removed from the mean. Any algorithm that wants to serve everyone fairly - relevantly, has to take this into account.