Shopping and Performance Max restructure for a scaling apparel brand
Blended ROAS improved from 3.1 to 4.6 over 90 days. Non-brand efficiency stabilized, waste dropped, and performance became easier to trust week to week.
3.1 to 4.6ROAS improvement
31% lowerWasted spend reduction
28% to 6%Brand leakage
Challenge
The account had overlapping Shopping and PMax coverage, weak branded demand separation, and inflated ROAS caused by brand-heavy traffic being absorbed into the wrong campaign types.
What changed
- Audited query mix across Search, Shopping, and PMax to isolate where branded demand was being misattributed
- Created a dedicated brand campaign and tightened negative-keyword controls across non-brand surfaces
- Rebuilt campaign structure around category logic and merchandising priority
- Introduced n-gram review to remove recurring waste patterns from search terms
- Split promotional inventory from core catalog so budgets could reflect business intent
The biggest issue was not simply that performance was uneven. The real problem was that the account structure made it hard to understand what was truly driving demand.
Once branded demand was isolated and category logic was restored, the account became easier to scale and easier to read. That improvement in clarity was as valuable as the direct efficiency gain.