Software / SaaS
SaaS vendor — PMax restructure
A SaaS vendor whose Performance Max spend was reporting success while quietly grinding budget into inventory nobody converted from. We restructured the asset groups and put the waste miner on PMax permanently.
Illustrative scenario — an anonymized composite of the engagement pattern we typically see, published until the client-approved case study lands. No performance figures are claimed.
The situation
A SaaS vendor had moved most of its Google Ads budget into Performance Max on the platform’s recommendation. The in-platform reporting showed engagement and conversions trending up — but trial signups from paid channels that the sales team could actually attribute had gone flat.
The classic PMax failure mode was in play: a catch-all asset group, loose audience signals, and a search-terms surface that mostly showed brand queries the company already owned.
What we did
The audit put structure first. Asset groups were rebuilt around distinct buying intents, audience signals were fed from real customer lists instead of guesses, and exclusions were set before budget was re-expanded. Every step was previewed, approved, and applied inside the same guarded loop we use for search.
- Brand-query cannibalization measured and fenced so PMax stopped reselling traffic the site already earned.
- A standing waste-miner pass on PMax search terms — irrelevant queries fenced with negatives and documented in the ledger.
- Creative decay watched weekly, with fatigued assets replaced through the normal plan-approve-apply cycle rather than left on autopilot.
What changed
Again, an illustrative narrative with no performance figures claimed. The meaningful change was epistemic: PMax stopped being a black box that reported its own success. Its spend now had to justify itself against evidence the client could read — in the search terms, in the ledger, and in the weekly graded plan.