-
Sacrificing to the Google Gods
Posted by on December 16, 2025 at 11:20 pmI can say with 100% certainty now that running a broad match experiment alongside my regular phrase match campaign improves performance.
I had always suspected this, but after I accidentally let my broad match experiment expire – my phrase match campaign tanked.
I gave the base campaign two weeks to see if it would improve on its own, but it did not and so I started another broad match experiment, and my base campaign is performing well again.
I have literally changed nothing other than having the experiment running. Same ads, same keywords. The only thing I do is keep up on the negative keywords.
And can absolutely 100% say without a doubt that my campaigns perform better with a broad match experiment running alongside the base campaign under phrase match. This is for my services based company using a target CPA and portfolio strategy to control max CPC (but still set very high – just to avoid the rogue $100 click on an avg cpc of $6-10).
I only ever set the experiment to use 20 to 25% of the budget. That is all taken into account and is still worth the cost. Just the sacrifice I need to make to satiate Google Gods.
Curious, if anybody else has come to this conclusion?
-
1 Reply
-
Sorry to break it to you but there’s no Google Gods. It’s just algorithms that have been programmed by people to respond to user inputs. Broad match + smart bidding has been overperforming everything else for at least 2 years now.
-
Sounds unreliable to me
-
There was an article I read that said the same thing. Broad match performs better than phrase and even at times exact.
I think it was a search engine land article -
Sounds like something a Google rep would say
-
All brand 😉
-
Yeah, a lot of us have seen this, even if Google will never admit it outright. Broad match seems to act like a signal miner for Smart Bidding, feeding the system query patterns and intent that end up helping the rest of the account perform better. As long as you cap it, keep negatives tight, and treat it like a tax for better learning, the “sacrifice to the Google gods” is often worth it.
-
Broad match includes all match types. So it makes sense the phrase match tanked. Yook into your search term report.
-
The reason why this is most likely misled is that your basing this on a feeling instead of data and facts.
Aside from the search terms Google doesn’t show us, you can actually go into the campaign/experiment and point exactly at the differences in performance and what actually led to that.
If you haven’t done that, you’re just operating on some whimsical gut feeling.
>I gave the base campaign two weeks to see if it would improve on its own, but it did not and so I started another broad match experiment, and my base campaign is performing well again.
Can you share some numbers? Are they even statistically significant?
Log in to reply.