Stomp the Google AdWords Expanded Broad Match Problem

As I previously blogged here, here and here, Google’s changes to its AdWords broad term matching algorithm have caused, for some advertisers, runaway spending without a commensurate increase in conversions.The Weasel

For months we tried running the excellent Search Query reports, finding bad matches, and adding corresponding negative keywords to ad groups to reduce the likelihood of future bad matches. But this was extremely time-consuming, and led to huge negative keyword lists. Worse, it was like playing “bop the weasel;” every time we ran a new Search Query report, we’d find additional new bad matches.

After much thought and discussion with other PPC mavens, we decided to take Matt Van Wagner‘s advice. We both observed that the EBM problem seemed to be most acute when the broad keyword contained fewer than three keywords. Since we rarely bid on one-word keywords, we simply stopped bidding on two-word broad match keywords.

Instead, we bid on the two phrase match variations of two-word keywords – for example, instead of bidding on the broad match version of the keyword ‘blue widgets,’ we bid on the two phrase match variations, “blue widgets” and “widgets blue.” This ensures we match for almost all relevant variations of the broad match equivalent. We’ll miss some matches we would have gotten through broad match – for example, neither of the phrase match versions will match the search term “blue sparkly widgets” – but in most cases that’s acceptable.

The result? Bad expanded matches disappeared and spending came back under control.

Comments (4)

  1. I’d love it if you and Matt could show me clearly the impact of this “ban the two-word broad match in favor of both major phrase matches” change – in terms of overall click and sales volume. Depending on the account, my sense is that the broad matches, and the expanded broad matching, are generating a lot of volume (albeit, not always good volume). Seems like it is a bit of a tightrope…

  2. Hey Andrew – my data is at a rather coarse level – Matt’s may be more refined: for the Clix client affected worst by EBM, ad spend increased over $7,000 the month we detected the issue. We spent a few weeks running Search Query reports and pouring negative keywords into the campaign, but every few days additional bad matches would pop up (“bad matches” as defined by search terms that bore little or no resemblance to the campaign keywords – but garnered clicks and few or no conversions). Once we employed the “Phrase Match Two-step,” the number of bad matches dropped precipitously, and spending and conversions came back in line with the previous month.

    I’ll email Matt and ask if he’s got more.

  3. I also encountered the same issue for one of my clients.
    In our case, the client’s brand term CPC shot up significantly because the brand term is triggered by irrelevant broad match keywords that are closely associated with the brand’s offering.

    I did pretty much the same thing as you did (ie. run regular Search Query report and add unwanted queries as negative keywords, more specifically exact negatives).

    What I also did additionally were to take the top 10% converting keywords of broad matches and re-introduce them into the campaign as exact matches. I did not use any phrase matches at all. This seem to work well too.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.