How to structure a Google Ads account when Performance Max is doing the heavy lifting

The problem with letting one campaign do everything

Performance Max is efficient at finding conversions wherever they are cheapest. Left unconstrained, the cheapest conversions in almost every account are people who were already going to buy — brand searchers and returning visitors. The campaign duly harvests them, reports an excellent return, and the business grows more slowly than the dashboard suggests it should.

The fix is not to avoid Performance Max. It is to remove its access to the easy conversions so it has to go and find new ones.

Separate brand into its own campaign, always

Brand search should sit in a dedicated Search campaign with exact and phrase match on your brand terms, its own modest budget and its own reporting line. Then add your brand terms as negatives to Performance Max through the brand exclusion list.

This single change usually makes acquisition costs look worse and the business look clearer. Reporting non-brand cost per acquisition separately is the only way to know what growth actually costs.

Tier the rest by intent, not by product category

A workable structure has three tiers. High-intent Search covers the queries that indicate a purchase decision — model numbers, 'buy', 'price', competitor comparisons. Performance Max handles discovery and shopping inventory. A third tier, usually Demand Gen or YouTube, does upper-funnel work with its own target and its own patience.

Splitting by product category instead produces campaigns that all target the same intent and compete for the same auctions, which is how budgets get spent without anyone being able to explain the marginal return.

Asset groups are the real lever inside PMax

Within Performance Max, asset groups are where you exert control: one per product theme, with its own creative, its own audience signals and its own listing group. A single asset group covering the whole catalogue gives the algorithm no structure to work with and gives you nothing to read in the reporting.

Audience signals are suggestions, not targeting. Feed them your customer match lists and high-value site audiences, then judge the campaign on new customer acquisition rather than on whether it obeyed.

Feed quality is a bidding input

For any account with Shopping inventory, the product feed does more for performance than bid strategy selection. Titles that lead with the terms people search, correct product types, GTINs where they exist, and images that meet requirements — these determine which auctions you enter at all.

We have seen feed title rewrites produce larger performance changes than any campaign restructure in the same account.

Measure new customers, not conversions

Set up new customer acquisition goals so the account reports how many buyers were genuinely new. Import offline conversions if your sales close on the phone. Set conversion values that reflect margin rather than revenue, if margins differ meaningfully across the catalogue.

Smart bidding will optimise faithfully towards whatever you tell it to value. Most disappointing accounts are optimising perfectly towards the wrong thing.

Every recommendation here comes from accounts we run. If something in your setup contradicts it, the account is the better authority — test it.

Where to go next

If this describes a problem you currently have, a short review of your accounts will identify whether it is the binding constraint or a symptom of something else. That conversation is free and does not require a commitment.