The problem in plain terms

Performance Max is designed to find conversions across all of Google's inventory — Search, Shopping, Display, YouTube, Gmail, Maps. Google's algorithm decides where to show your ads and how to allocate budget. In theory, it finds the highest-value opportunities. In practice, for most eCommerce brands, it does something much simpler: it follows the path of least resistance.

The path of least resistance is branded search. People who are already looking for your brand by name. They were going to find you anyway. They convert easily, the CPA looks great, the ROAS looks strong. Performance Max takes credit for them, your reports look healthy, and meanwhile your non-brand acquisition — the actual growth engine — is being systematically underfunded.

How to diagnose it

The clearest signal is the gap between your reported ROAS and your actual new customer acquisition rate. If ROAS looks strong but your revenue isn't growing proportionally, brand cannibalisation is almost always part of the story.

More specifically, here's what to look for in the account:

• In your Search Terms report (Insights → Search Categories), look at what percentage of PMax impressions and clicks are on branded terms. If it's more than 30–40%, that's a red flag.
• Compare your branded Search impression share before and after launching PMax. A significant drop in your standalone brand campaign's impression share suggests PMax is competing for the same traffic.
• Look at your new vs returning customer split in your analytics. If PMax is responsible for most conversions but your new customer percentage is declining, the math doesn't add up.

Why Google's suggested fix doesn't fully work

Google will tell you to add brand terms as negative keywords at the campaign level, or to use brand exclusions in the PMax settings. This helps — but it's not a complete solution. PMax can still serve on branded queries through the Shopping and Display inventory where keyword-level negatives don't apply. And Google's brand exclusion tool is not granular enough to handle complex brand term structures.

The more reliable fix involves restructuring how you handle brand vs non-brand at the campaign architecture level — not just applying negatives and hoping.

---

The fix: separate brand from non-brand at the architecture level

The structural solution has three components working together.

1. Dedicated brand Search campaign

Run a separate, tightly controlled Search campaign for all branded terms — your brand name, common misspellings, brand + product combinations. Set this campaign to Maximise Clicks or a manual CPC strategy. The goal is to own your brand traffic efficiently, not to maximise ROAS. Keep the budget modest — branded clicks are cheap and highly efficient already.

2. Brand exclusions on PMax

Apply brand exclusions in your PMax campaign settings to prevent it bidding on your core branded terms. This won't catch everything (see above), but it significantly reduces the overlap. Pair this with campaign-level negative keyword lists where possible.

3. Segment PMax by intent

Instead of running one PMax campaign across everything, structure asset groups to reflect different types of buyer intent — new customer prospecting, category-level shopping intent, retargeting. This gives the algorithm clearer signals and makes it easier to see where budget is actually going.

What to expect after restructuring

In the short term, your reported ROAS will likely drop. This is not a bad sign — it's PMax losing access to the easy branded conversions it was inflating your numbers with. What you should see over 4–8 weeks is a meaningful increase in new customer revenue as the non-brand budget actually reaches people who haven't bought from you before.

Track new vs returning customer revenue separately. That's the metric that tells you whether the restructure is working, not blended ROAS.