$80k → $420k/mo

The brand had been stuck at $80k a month on Meta for a year. Every push above that line collapsed ROAS inside 72 hours. Three agencies had taken a run at the ceiling. All three blamed the creative, and all three were wrong.

The account structure was the problem. 34 campaigns, most launched by previous agencies, most never turned off. Meta’s auction treats overlapping audiences as competition, so every new spend dollar had to outbid the brand’s own legacy campaigns before it reached a customer. The algorithm never accumulated enough signal in one place to exit the learning phase cleanly. Inside that structure, creative quality was irrelevant. Better ads would have lost the same auction.

Week one was demolition. We consolidated 34 campaigns into 7: four ASC and three manual prospecting sets targeting specific retention cohorts. Spend initially dropped because Meta was relearning, and ROAS wobbled for eight days. The founder was nervous. Day nine the curve flipped.

The second fix was cadence. Most agencies in this space run creative tests in three-week sprints. We ran them daily. Two new ad variants ship every weekday. Winners auto-promote on day three, losers shut off. No 90-minute creative-review calls. About 130 variants shipped in the first quarter, 3.5× the brand’s prior cadence.

The third fix was measurement. Day-1 platform ROAS punishes exactly the moves that make scaling work, so we moved decisioning to 14-day blended contribution, pulled nightly from the warehouse. The founder got a number that matched the P&L instead of Ads Manager, which made each budget increase a calculation instead of a leap.

By month six, Meta spend had climbed 5.2× off the starting point, from $80k to $420k a month. Blended CAC held flat. At that spend level, flat is the goal. Revenue followed spend on the clean line every operator wants to see.