Performance Max beat branded search on CPA. That is not what the playbooks say.

Performance Max earned its early reputation. Targeting was opaque, attribution was worse, and it absorbed budget while telling you very little about where it had gone. For a long stretch the sensible advice was to keep it small or keep it out.

That advice is now out of date, and holding onto it is costing advertisers real money.

The two changes that mattered

Negative keywords in Performance Max. This addressed the largest single objection. Excluding terms gave advertisers an additional layer of control over irrelevant Search and Shopping queries, which helped reduce some unproductive clicks and spend. It is not the granularity of a Search campaign, and it is not complete control over targeting — but it is enough to run the campaign responsibly.

Holistic journey attribution. Newer attribution reporting makes it possible to understand the customer journey through Performance Max rather than inferring it from what changed elsewhere. That is the difference between a campaign you can optimise and a campaign you can only switch on and off.

Together they changed what the campaign type is for.

Where it now fits

Our read, after scaling it since those changes:

The share of revenue reflects that shift: in accounts our team has scaled since negative keywords and the new attribution reporting arrived, roughly 40% of Google Ads-attributed revenue came through Performance Max. That is an attribution figure, not an incrementality claim — it describes where the platform assigns credit, not what would have happened without the campaign.

The finding that surprised us

We ran an incrementality test comparing campaign types on branded search. Performance Max came out ahead on CPA.

That runs against the conventional structure, which treats branded search as a tightly controlled Search campaign and keeps Performance Max away from brand terms entirely — usually with brand exclusions applied specifically to prevent it from claiming that traffic.

We are not arguing the conventional structure is wrong. We are arguing it is an assumption, and assumptions in this part of the account are worth testing rather than inheriting. The result came from a test we could have skipped, on a question we thought was settled.

This was the result of our test, not a universal rule — which is why we think this kind of assumption is worth testing in the account itself. Test it in your own before acting on ours.

Coordination is the part people skip

Performance Max running alongside Search is not the same as Performance Max coordinated with Search, and the gap matters most when Search is running an expensive strategy.

If Search is on value-based bidding — selective, deliberately conceding volume in exchange for lead value — then what Performance Max is doing with the traffic Search declined is a strategic question, not an afterthought. Left uncoordinated, the two campaign types compete for the same auctions, attribution muddies, and you end up unable to explain your own results.

Set the division of labour deliberately. Decide which campaign type owns which part of the funnel, apply exclusions to enforce it, and review the boundary as the account matures.

The short version

• Negative keywords and journey attribution addressed the two objections that justified avoiding Performance Max.
• In the accounts we compared it worked better than Display remarketing for retargeting, and better than Demand Gen for discovery — the right mix still depends on the account.
• In accounts scaled since those changes, roughly 40% of Google Ads-attributed revenue came through it — an attribution figure, not an incrementality claim.
• An incrementality test showed it beating traditional branded search on CPA — worth testing in your account, not worth assuming.
• Coordinate it with Search deliberately, especially when Search is running value-based bidding.