Performance Max isn’t the problem most advertisers think it is. The problem is almost always one of three things: the campaign was launched with no exclusions or feed cleanup, the budget was cut or changed too aggressively for the algorithm to keep learning, or PMax was left to run in isolation with no standard Search campaign protecting the branded terms that were already converting. Each one is fixable, and none of them require abandoning Performance Max entirely.
The budget-change mistake that resets everything
Cutting a Performance Max budget by more than 20% in a single change resets its learning period — the campaign effectively has to relearn what’s working from a reduced data set. The fix is mechanical: if spend needs to come down, do it in 10-15% steps spaced five to seven days apart, not one large cut. Premature changes and learning-phase resets like this are estimated to waste 15-25% of monthly ad spend on their own, before any other mistake is even in play.
The instinct to react fast to a bad week is usually the thing making the bad week worse.
How much budget PMax should actually get
There’s no universal number, but most accounts land somewhere between 25% and 40% of total Google Ads budget in Performance Max, with that range shifting based on how much of your conversion volume comes from Shopping-eligible products versus lead generation. Accounts with lower overall conversion volume sometimes see better cost-per-acquisition keeping PMax closer to 10-25% of budget — a smaller, well-fed campaign frequently outperforms a larger one that’s data-starved relative to its scope.
Two failure modes sit on either side of that range: starving PMax with too little budget to ever learn effectively, and flooding it with budget and no constraints, which tends to push spend into low-intent Display and YouTube placements the algorithm defaults to when it lacks better signal.
Search themes and negative keywords still matter in PMax
Performance Max is automated, not unsteerable. Search themes and negative keywords are the two levers that control which queries the campaign enters the auction for — used well, they steer the algorithm toward high-value intent and block obviously wasteful spend without undermining its ability to discover converting patterns you wouldn’t have thought to target manually.
Don’t run PMax alone
Running Performance Max in isolation, with no standard Search campaign alongside it, is increasingly treated as a structural mistake rather than a minor inefficiency. PMax can end up competing with — and sometimes cannibalizing — branded search terms that were already converting reliably before it launched. A geo-split or brand-exclusion test is the standard way to check whether that’s happening in a specific account rather than guessing.
One more default worth checking today
New campaigns are opted into Google’s Search Partners network by default, which includes lower-quality parked domains alongside legitimate sites like Amazon. Traffic from that network can carry a cost-per-conversion 33-50% higher than Google.com traffic at the same bid. Unchecking “Include Google Search Partners” is a five-second fix with an outsized effect on accounts that have never touched it.
FAQ
Is Performance Max worth using for a small budget account? Yes, but expect a longer learning period and keep it to a smaller share of total spend — 10-25% rather than 40% — until conversion volume is high enough to feed it properly.
How do I know if PMax is cannibalizing my brand search instead of adding incremental conversions? Run a geo-split test: pause PMax in a subset of regions while keeping brand Search live, then compare conversion volume against regions where both are running. A meaningful drop in the paused regions means PMax was adding real volume, not just capturing clicks that would’ve converted anyway.
Ratnawat Global manages Google Ads and Performance Max accounts built around exactly this kind of budget discipline — not just campaign setup. Get in touch if your account’s been running on autopilot longer than you’d like to admit.