A client, who happens to be in a very competitive search space, was insistent that we run a very high-level keyword on broad match, as a result they were paying extremely high CPCs on all sorts of long tail queries.  With “widgets” running on broad match, they were paying $10-20 for queries like “widgets in sioux falls south dakota” and “bob smith’s custom widgets.”

We quickly realized that the standard approach of reactively adding negative keywords based on search term reports wasn’t working – there were simply too many geographic, proper name and other modifiers.  Chasing them one-by-one was a losing proposition, and every day brought new long tail queries at $10-20 per click.

What to do?  Turning off the broad match keyword wasn’t an option – not only did the client insist that it stay live, but there were legitimate reasons to keep it.  Most of the modifiers were names or places, so here’s what we came up with:

1. Grab lists of the 1000 most populous US cities and all 50 states, and of the most popular first & last names…both easily found on the Web.
2. Add all of them as negative keywords to the campaign containing the high-level, broad match keywords.
3. Then, create a new campaign that includes the keyword plus each city/state/name (“widgets in topeka kansas,” etc.) and obvious variations, at much-reduced Max CPCs.  Since there were so many long-tail possibilities, we started these on Broad Match to make sure we weren’t sacrificing volume.
4. Keep watching Search Term Reports in both the original and new campaigns.

So far so good…we’re still evaluating the net effect on costs, clicks and conversions, but there’s a LOT less garbage to sift through in search term reports, and I haven’t had one of those “Oh God, I can’t believe we paid THAT MUCH for THAT KEYWORD” moments in over a week!