−91% campaigns, +2.1× revenue
the learning phase finally unlocked
Results from specific engagements. Every account starts with an audit because outcomes depend on your product, margins, and creative.
01The symptom
The account had 47 active Meta campaigns, most delivering under $5/day. The previous agency had been running a 'test everything' strategy for 14 months. Meta couldn't exit the learning phase. Scale was impossible.
02The diagnosis
The previous agency's 'test everything' strategy split conversion volume across 47 campaigns, so nothing could reach the 50 conversions per week per ad set Meta needs to exit the learning phase. The ads were not failing. The structure guaranteed they never got the chance to work.
03What we did
- Full account teardown — 47 campaigns → 4, structured around ASC + two high-budget manual prospecting sets
- Migrated catalog ads to advantage+ shopping with dynamic exclusion rules against purchasers
- Built a creative scorecard so the brand stopped shipping lookalike versions of dead ads
“Revenue went 2.1× in ninety days and we pay a third less for a customer than we did last year. Fourth agency in two years — the first where the results showed up in the bank account.”
04The full story
We were the fourth agency on this account in two years. The previous one had run a “test everything” strategy for 14 months, and the account showed it: 47 active campaigns, most delivering under $5 a day, none of them able to scale.
The diagnosis was in the delivery data. Meta needs 50 conversions per week per ad set to exit the learning phase. This brand’s conversion volume was being split 47 ways, so nothing exited learning, ever. Fourteen months of “testing” had produced no learnings because no test ever reached significance. The previous agency had been reading noise and calling it strategy.
Week one, we asked for COGS, returns data, and the cohort table before touching a single campaign. You cannot set a real CAC target without them, and nobody had asked for them before.
Then we turned things off. 47 campaigns consolidated into 4: one broad ASC, one retargeting, two manual prospecting sets against behavioral segments. Dynamic product ads got rebuilt against the full catalog with proper exclusion rules; the old setup was showing customers products they had bought the week before. Spend concentrated into fewer ad sets, and conversions per set climbed above the 50-a-week threshold inside two weeks.
The last piece was a creative scorecard. Part of the bloat came from shipping incremental versions of ads that were already dead, on the theory that the creative just needed one more iteration. The scorecard ended that habit. An ad either cleared its minimum threshold in the first 48 hours or it did not get a variant. No exceptions, no debate.
Ninety days in: 91% fewer campaigns, revenue up 2.1×, CAC down 34%. None of the work was clever. Most Meta accounts are broken for boring reasons, and the fix is discipline applied early and held daily.