Google Performance Max Attribution: What It Hides and How to Measure It

Performance Max is Google's fastest-growing campaign type and its most opaque. Google automates everything — placements, audiences, bidding, creative combinations — and reports back a ROAS that sounds great and tells you almost nothing.

If you're running PMax and making budget decisions based on its reported numbers, you need to read this.

What Performance Max Actually Is

PMax is a single campaign type that runs across every Google property: Search, Shopping, Display, YouTube, Discover, Gmail, and Maps. Google's algorithm decides in real-time which placement, audience, and creative to use for each impression.

The pitch: give Google your assets and your conversion goals, and the algorithm will find customers wherever they are most efficiently.

The reality: Google gains control over where your money goes, and you lose the ability to audit it at any meaningful level.

The Four Attribution Problems With Performance Max

1. It cannibalizes your existing channels

PMax serves on branded search terms — your own brand name, your own product names. When someone Googles "YourBrand pricing" and your PMax campaign intercepts that click, Google reports it as a PMax conversion. But that customer already knew your brand. They were going to convert anyway.

This is called cannibalization, and PMax is particularly aggressive about it because branded search terms are high-intent and easy to convert — exactly what the algorithm optimizes for.

How to check: Compare branded search volume and branded ROAS in your standard Search campaigns. If branded query volume drops after launching PMax, PMax is stealing those conversions.

2. Cross-channel attribution is hidden

PMax doesn't tell you which placements or channels drove which conversions. You can see that PMax got 120 conversions, but not how many came from YouTube vs. Display vs. Shopping.

This makes it impossible to know whether PMax's Display component is wasting money, whether YouTube is contributing any real conversions, or whether Shopping alone would perform the same.

Google says they'll improve reporting. They've been saying that since PMax launched.

3. Reported ROAS includes view-through conversions

Google's default PMax reporting counts view-through conversions — people who saw a Display or YouTube ad but never clicked, then converted later through any channel.

A customer who ignored your YouTube ad and converted via Google Search a week later can show up as a PMax conversion. Your PMax ROAS includes these. Your actual PMax ROAS does not.

How to fix: In Google Ads Conversion settings, set view-through conversion window to 0 days. This removes the noise.

4. Smart bidding biases toward easy wins

PMax uses smart bidding, which optimizes for conversion volume at a target ROAS or CPA. The algorithm naturally gravitates toward high-intent, low-funnel users — people who were going to convert anyway.

This inflates ROAS in the short term while potentially starving upper-funnel investments that would generate new demand. You get great numbers today and a shrinking new customer pipeline next quarter.

How to Actually Measure PMax Impact

Step 1: Create a brand exclusion list. Go to Tools → Shared Library → Brand Exclusions and block your own brand terms from PMax. This forces PMax to compete on non-branded terms and gives you cleaner incrementality data.

Step 2: Turn off view-through conversions. Remove view-through attribution from your conversion actions so PMax only gets credit for actual clicks.

Step 3: Run a holdout test. The only true test of PMax incrementality is to pause it for 2–4 weeks and observe what happens to total revenue. If revenue barely moves, PMax was taking credit for conversions that would have happened anyway.

Step 4: Compare to actual revenue. Pull new customers from Stripe or Shopify for the period. Compare to PMax's reported conversions. The gap is the overcounting.

Step 5: Track blended ROAS separately. Don't evaluate PMax in isolation. Your true measure is: did total revenue grow in proportion to total spend after PMax launched? If total spend went up 30% and revenue went up 12%, something is wrong — even if PMax reports a great ROAS.

The Bottom Line on PMax

Performance Max can work well — particularly for Shopping-heavy campaigns at scale. But its default reporting is optimized to show you numbers you want to see, not numbers that reflect reality.

Treat PMax ROAS the same way you'd treat any platform-reported number: directionally interesting, not operationally reliable. The only number worth budgeting on is your true blended ROAS from your revenue source.

See how PMax performance fits into your real ROAS →