Performance Max: Complete Setup and Optimization Guide
Performance Max campaigns use Google's full inventory. Done right, exceptional ROAS. Done wrong, budget burn with no insight. Here's how to do it right.
Performance Max takes the controls and asks for trust. Sometimes it earns it. The problem is you can't see why.
It runs simultaneously across Search, Shopping, YouTube, Display, Discover and Gmail, on one budget with one goal. You supply the assets and the signals, Google picks the rest.
When it works
E-commerce with a good feed. This is where it's strongest, because Shopping does the heavy lifting and the other channels fill in.
Enough conversion volume. Below ~30 conversions a month per campaign, the algorithm has nothing to learn from and spends badly.
Clean conversion data. If tracking is broken, PMax optimises towards the wrong signal faster than a manual campaign would.
When it doesn't
Small budgets. Spread across six channels, they reach nowhere.
Lead generation with variable lead quality. PMax optimises towards completed forms. If half of them are junk, it will bring you more exactly like them.
The fix: send offline conversions back — which lead became a customer — not just the form submission.
What you actually control
Asset groups. Split them by theme, not in one pile. One group per product category or service.
Audience signals. They aren't targeting, they're starting hints. Give it your customer lists — that's what it learns from best.
Brand exclusions. Without them, PMax cannibalises searches for your own name, which would have come anyway, and shows you a beautiful, false cost per acquisition.
The feed. In e-commerce, feed quality matters more than any campaign setting.
What you can't see
Reports won't tell you how much went to each channel, which placements ate the budget, or which terms triggered impressions. Scripts exist that extract more, but transparency stays far below a classic campaign.
How to test it honestly
Don't compare PMax to Search in an account where both bid on the same traffic. Run one period with it off and one with it on, at constant total budget, and look at total account conversions, not campaign-reported ones.
It's the only test cannibalisation can't fool.
How many hours a week do you lose to repetitive work?
In 20 minutes on Zoom we work out which of them can run on their own. You leave with the answer, whether we work together or not.
Free call, no obligations. You leave with a concrete list of problems and automation opportunities, whether or not you choose to work with us.