For example, if broad match keyword "widget" performs well and earns a high bid, searches for "red widget" will trigger the "widget" ad instead of the "red widget" ad. Because of this, the landing page is non-optimal and conversion info that should belong to the "red widget" keyword ends up with the "widget" keyword.
Has anyone found a way to counteract this?
The only thing I can think of is to add negative keywords to the ad group containing the problematic broad match keyword. The negative keywords would match the more specific keywords which are being stolen from. "-red widgets" would be an example.
The only problem with this is scaling. I have about 4000 ad groups managed automatically via the API, each containing a different keyword. Will Google allow me to add 4000 negative keywords to each of my 4000 ad groups?
Quality score is determined by many factors, yet is calculated when the user's query matches your keyword.
Therefore, if you have a keyword in the same ad group that is exact and broad matched, they should have the same quality score.
You can therefore set a higher bid for the exact match (usually 20% + difference is enough) than the broad match and the exact match keyword should be shown instead of the broad match when either keyword is can be triggered by the search query.
But it happens consistently, not a very high volume but it still happens. For a system built on rules, you would think it would never happen, but it does.
I run all my exact and phrase terms as negatives in the Broad adgroups (or campaigns) as others have suggested. So far I haven't hit a limit. I currently have 50k negatives in my account, and 8k the most in a single campaign.
I've found that I have to keep a close eye on broad match adgroups or else they eat up all my optimizations to the rest of the account.
What happens is there are variations of the broad match that you are being matched to that are converting.
For instance, let's assume you bid on the broadmatch version of widgets. You will find that your broadmatched word was actually shown for these searches, and let's assume this is the corresponding Conversion Rate (CR):
widgets 1% CR
red widgets 3% CR
yellow widgets 10% CR
pink widgets 0% CR
red orange yellow widgets 22% CR
red orange yellow pink widgets 0% CR
To learn this information, run a search query report.
Then, add pink as a negative keyword, add the higher CR variations as positive exact/phrase keywords.