Brand traffic inside Performance Max: the hidden subsidy
The most common PMax illusion I open in audits isn't junk placements — it's brand. Someone hears about your clinic from a friend, googles your name, clicks the shiny ad above your own organic listing, and PMax books an acquisition. The campaign didn't find that patient. It taxed them. And because the interface aggregates everything, owners see a blended number that looks like the machine is printing money.
On this page
How PMax captures your brand demand
PMax targets by signals and assets, not keyword lists, and unless told otherwise it treats your brand name as just another high-converting query — the highest-converting one, since those searchers already decided. Branded clicks are cheap, convert at multiples of cold traffic, and pull the campaign's blended CPA down to a number no cold channel could produce. That blend is the subsidy.
Measuring the brand share
Google now shows more than it used to: search themes and search-term insights at campaign level reveal the query categories PMax served on, and brand terms are usually sitting near the top by conversions. Cross-check from the other side: your brand campaign's (or organic's) impression trend often dips as PMax ramps — demand didn't grow, it moved. Put a number on it: what share of PMax conversions came from brand-category terms. In accounts I open, finding a third or more is unremarkable.
Excluding brand the right way
Brand exclusions exist at the campaign level for PMax — apply your brand terms and common misspellings there, and keep a dedicated brand Search campaign to catch that demand deliberately, cheaply, with copy you control. Do it in that order: exclusion without a brand campaign hands your name to competitors bidding on it. Account-level negative keywords now also apply to PMax, which helps enforce the boundary consistently.
Reading the account after exclusion
Expect PMax's reported CPA to rise and volume to fall — that's not damage, that's honesty. What you're seeing for the first time is the campaign's real cost of acquiring someone who hadn't already chosen you. Judge it on that basis against your other cold channels, using CRM-verified outcomes, and only then decide whether PMax deserves its budget. A falling blended CPA was never the goal; a true cost per new patient is.
When keeping brand in PMax is defensible
Almost never for high-ticket clinics — but there are edge cases: tiny brand volume where a separate campaign is overhead, or heavy competitor pressure on your name where PMax's asset coverage defends more surfaces. If you keep it, keep it knowingly: segment reporting so brand and non-brand are never read as one number.
VERIFIED EXAMPLE · When I inherited a wasteful account at a luxury US healthcare clinic, it was optimizing toward platform conversions. Rebuilt around admitted patients imported from the CRM, cost per admitted patient fell ~78% while spend scaled ~2× and approved patients rose ~9×.
| Metric | With brand inside PMax | After excluding brand |
|---|---|---|
| Reported CPA | Flattered by cheap branded clicks | Rises — now honest |
| Reported volume | Inflated by patients who already chose you | Falls to genuinely new demand |
| What you're actually measuring | A blend of new and existing demand | The true cost of acquiring someone new |
Questions owners ask
Our PMax numbers are great. Why would I break that?
If a meaningful share is brand, the numbers aren't great — they're blended. You're paying the platform for patients who typed your name. Excluding brand doesn't lose those patients; a brand Search campaign catches them at a fraction of the cost.
Will excluding brand hurt overall conversions?
Total conversions shouldn't fall — they redistribute to the brand campaign and organic. What changes is attribution honesty: PMax keeps only what it genuinely found.
How do I check this without an agency?
Open PMax search-term insights and look at the top categories by conversions. If your clinic's name is there, you have your answer. The deeper channel-split analysis takes scripts, but the brand check takes ten minutes.