PMax vs Search for high-ticket: when the restructure pays off
Performance Max is neither the scam its critics describe nor the default its advocates assume. It is a bidding system that amplifies whatever signal you give it, across inventory you cannot fully see.
That combination is dangerous in high-ticket, where conversion volume is low and the gap between a lead and a customer is enormous. It is also, configured properly, where a lot of incremental volume lives.
On this page
What PMax actually changes
It hands campaign structure to the algorithm and reduces what you can observe and control.
One campaign spans Search, Shopping, Display, YouTube, Discover, Gmail and Maps. You supply assets, audience signals and a budget; Google decides where impressions go. Reporting arrives at asset-group level, and while search terms visibility has improved, you still see less than you would in a Search campaign.
None of this is inherently bad. It is a trade: less control and less visibility in exchange for reach across inventory you would otherwise have to manage separately. Whether that trade is good depends entirely on whether the algorithm is optimizing toward something real.
Three conditions where it works
All three, not two of three.
| Condition | Why it is non-negotiable |
|---|---|
| Conversion signal reflects revenue | PMax is more dependent on signal quality than any other campaign type, because you cannot manually correct its targeting. Fed form fills, it optimizes toward form fills across far more inventory. |
| Brand excluded | Without brand exclusions, PMax captures people searching your name and reports them as acquisition. Performance looks excellent and nothing incremental happened. |
| Enough conversion volume to learn | A campaign producing a handful of confirmed customers a month gives the model very little. Below that threshold, tightly managed Search usually beats it. |
Where all three hold, PMax genuinely finds demand that keyword targeting misses — particularly on YouTube and Discover, where high-ticket healthcare buyers research long before they search.
How it burns high-ticket budgets
Three failure modes, in descending order of frequency.
Brand cannibalization. The most common by far. PMax absorbs branded searches, records them as conversions, and reports a cost per acquisition that is the envy of the account. Budget shifts toward it, real prospecting shrinks, and total customers fall while the dashboard improves.
Signal amplification. If a page-view conversion or a 60-second call is in your goal set, PMax will find enormous quantities of both. In a Search campaign you would notice the bad queries. Here you often cannot.
Inventory drift. Without exclusions, spend leaks toward cheap Display and app placements that produce clicks and no consultations. Placement exclusion lists and content suitability settings exist for this and are frequently left at defaults.
Testing the split honestly
The comparison most accounts run is invalid, because PMax is quietly being fed traffic Search would have won anyway.
- Apply brand exclusions to PMax first. Every number before this is contaminated. This step alone changes the conclusion in most accounts.
- Give both the same conversion goal — the CRM-verified outcome, not form fills. Comparing campaigns judged on different definitions tells you nothing.
- Run for a full sales cycle plus 30 days. In high-ticket that is often 90 days. A 30-day test measures the learning period.
- Compare cost per acquired customer, not cost per conversion, not ROAS on modeled values.
- Check incrementality where you can. If pausing PMax for two weeks leaves total customers unchanged, you learned the most useful thing available.
Then decide per service line rather than for the account. It is common for PMax to win on a high-volume mid-ticket procedure and lose badly on the flagship one.
A note on falling click-through rate
When an account moves from Search-led to PMax-led, click-through rate usually falls. This is expected and is not a regression.
Search ads answer an explicit query, so a high proportion of people who see them click. Display, YouTube and Discover placements reach people who were not searching at all, so the same number of customers arrives alongside far more impressions. The ratio drops because the denominator changed.
Judging the restructure on that ratio will lead you to reverse a change that was working. The metric to hold onto is cost per acquired customer, confirmed in the CRM — which is the only one that survives a change in channel mix.
Google's Performance Max documentation covers the settings; brand exclusions are configured at campaign level and are the first thing to check in any inherited account.
VERIFIED EXAMPLE · When I inherited a wasteful account at a luxury US healthcare clinic, the restructure was validated on CRM-confirmed admissions rather than platform conversions — which is what allowed spend to scale ~2× profitably instead of being cut.
Questions owners ask
Is Performance Max bad for high-ticket businesses?
Not inherently, but it is unforgiving. It amplifies whatever conversion signal it is given across inventory you cannot fully inspect, so a weak signal does more damage here than in a Search campaign. With CRM-verified conversions and brand exclusions in place it can find genuine incremental demand.
What are brand exclusions and why do they matter so much?
They stop Performance Max serving on searches for your own business name. Without them PMax intercepts people who had already decided to contact you and reports them as new acquisition, which makes its performance look excellent while nothing incremental happens. It is the first setting to check in any inherited account.
How long should a PMax test run?
A full sales cycle plus about 30 days, which in high-ticket healthcare usually means 90. Shorter tests mostly measure the learning period, and the learning period is not what you are trying to evaluate.
Why did my click-through rate drop after moving to PMax?
Because the denominator changed. Search ads answer an explicit query and get clicked often; Display, YouTube and Discover reach people who were not searching, producing far more impressions for the same customers. The drop is a structural artifact, not a performance regression.
Should I run PMax and Search at the same time?
Usually yes, with clear separation: brand handled in its own Search campaign and excluded from PMax, high-intent keywords in Search where you can control them, PMax for discovery. Decide the balance per service line rather than for the whole account.