-
Google Ads phrase match close variants spiraling out of control – anyone else seeing this?
Posted by on November 11, 2025 at 6:05 pmI help run Google paid search campaigns for a B2B SaaS company with a pretty sizable account ($1.4M/month). Search has always been our best performer, and the vast majority of our keywords are phrase match. But over the past year, I've watched Google get increasingly aggressive with phrase match (close variants), and it's starting to become a real problem.
The numbers are concerning. Phrase match close variants are now making up almost 76% of our traffic, with most being irrelevant, poor-intent matches. CPCs for these close variants have jumped nearly 200% from last year and are now almost exactly the same as exact match.
Other things I've noticed:
- Huge spike in competitor branded search terms matching with our unbranded product keywords (in some cases, competitor terms are getting more click volume and spend as partial matches than in their respective dedicated competitive campaigns)
- Keywords matching with search terms for completely different products that we own
- Specific long-tail keywords matching with overly broad 1-word terms
- Major increase in educational and navigational search terms that don't match our keyword intent at all
We've been adding negatives aggressively, but with an account this size, it's like playing whack-a-mole with 50,000 moles. The volume is just too high to manage manually long term.
I'm considering two approaches:
- Building a weekly script to automatically flag trending irrelevant terms
- Shifting more budget to exact match (though that's not ideal since it would be extremely time-consuming to scale, and who knows what Google will do next)
Has anyone else run into this? Do we understand what's driving this change on Google's end? And more importantly, has anyone found a better way to manage this at scale?
Would love to hear from other folks managing large B2B search accounts especially.
-
1 Reply
-
GAds have beein going down this way for quiet some time now. Even exact match is showing this.
With this amount of traffic wouldn’t it make sense to consider going pmax? It has never worked for me in b2b, but we always thought that would be because of the amount of conversion signals we would be able to send.
-
Google has been playing a lot more fast and loose lately. It is good you are doing the work with all the negative keywords.
Some friends have tried to automate this SQR work with tons of false positives as the outcome. As long as someone is shifting through what the report spits out then I think some automation can help take the load off.
-
Run a search term exclusion script that flags low intent variants by regex on cpc and ctr thresholds then feed negatives weekly while moving high value queries to exact in a shared list
-
Yes but make sure to review search terms in order to combat negative terms.
-
You should focus on revising your keyword strategy into exact and broad match options. Exact match will, obviously, continue to give you fairly close matching queries for greater control.
These days broad and phrase can serve for almost the same range of queries. Yes, the matching rules for phrase are tighter, but that’s not the whole story.
Broad match will use a variety of additional signals including other keywords in the ad group, landing page information, and some user characteristics to bid more efficiently and for more relevant queries. This often more the compensates for matching rules.
Plus, only broad match (or keywordless) qualifies for placement inside of AI overview and AI mode results. So it’ll be important to have a solution in place to address this as adoption continues to grow.
-
Matching for us hasn’t been that bad. We run full broad across all search and have been for about a year now. Nothing too out of control except competitor terms – whole reason why everyone’s brand terms are up 10x in price over the last 1-2 years (blame PMax as well for this). Our branded search campaign runs us almost $700/day now compared to $150 or so.
Log in to reply.