Performance Max for eCommerce
Performance Max isn't the problem. The problem is feeding it the wrong signals.
If PMax optimises against add-to-carts, duplicated values, a dirty feed or low-margin products, it can scale pretty numbers and weak commercial results.
If PMax optimises against add-to-carts, duplicated values, a dirty feed or low-margin products, it can scale pretty numbers and weak commercial results.
What to check
- Purchase as the primary conversion.
- Order value that isn't duplicated.
- A complete Merchant Center feed.
- Custom labels for margins, brands and seasonality.
- Brand and non-brand read separately.
- Products that absorb the budget.
When to use it
- You have enough conversions.
- The catalogue is segmented.
- Tracking and Consent Mode are correct.
- You can read margins, returns and real orders.
When to be careful
- A small budget against a large catalogue.
- A feed with no custom labels.
- ROAS calculated on gross revenue alone.
- One campaign with everything inside it.
How we run it
Before the campaign
We verify purchase, value, currency and transaction ID; we check the feed, destinations, availability and Merchant Center policies. If the starting data is inconsistent, fixing it comes before increasing spend.
During optimisation
We annotate changes to budget, goals and segmentation, and leave a reasonable window before judging them. We compare attributed data with orders, returns and margins where available, without treating platform ROAS as profit.
Alternatives and tests
Performance Max is neither mandatory nor right for every account. Standard Shopping, Search or a different structure can offer more control. The choice depends on volume, catalogue, demand, budget and measurement quality.
What we don't promise: an automated campaign does not guarantee orders, ROAS or growth. We document assumptions and success criteria before testing, so a negative result stays informative instead of being hidden behind aggregate conversions.