I know there's a process and algorithm in place to detect bad clicks but it just doesnt seem enough. So I'm trying to prove my case. I have tried things like using analytics to count the average time on site for the dubious clicks - figuring these wont be too long if they're fake.
Does anyone have any ideas of tell tale signs for bad adwords or any further analysis I can do?
Are you able to track the source of the bad clicks to a particular address?
I discovered that I am getting thousands of impressions from Gmail, YouTube & MySpace. None which, by the way resulted in many clicks & no conversions.
Using site exclusion I blocked YouTube & MySpace while leaving Gmail. That cut the daily number of impressions back quite a bit with no impact on clicks. I am still balancing whether to block Gmail.
I've been using the AW option to optimise for conversion cost, which means that G could/should automatically weed out bad sources (amongst other things) for you that you'd never even have the data for.
That makes the content network a safer place for advertisers IMHO.
So far, using that and (separately) PPA, I've reduced conversion cost by about 20% albeit on small budgets.
Rgds
Damon