The numbers said it was working. The business said otherwise.

A DTC dog food subscription brand. Strong product-market fit, loyal base, and genuine traction - mid-seven figures, mostly Meta-driven. The kind of brand where the founder could tell you exactly why the kibble was better, and be right.

For two years, the Meta dashboard had told a story of success. ROAS looked healthy. Cost-per-acquisition was 'within target.' But every time they pushed spend higher, they somehow ended up making less money. Not dramatically - just enough to be confusing. They'd run through two media buyers trying to solve it.

Figure out why scaling was making the business worse. Build a version of growth that actually worked.

Started by pulling the real numbers, not the dashboard numbers. Post-purchase surveys, cohort tracking, incrementality testing. What we found: two of their top three campaigns were mostly taking credit for customers who would have signed up anyway. The 'performance' was real - the attribution wasn't. Restructured spend around audiences that were genuinely new to the brand.

• True CAC dropped 31% once cannibalised organic demand was removed from the picture
• Scaled monthly spend 65% while staying inside target unit economics
• Month-3 subscription retention improved 18 points - new audiences actually stayed, not just trialled
• Finance and marketing looked at the same numbers for the first time

"Honestly, it was a bit humbling. We thought we were good at this. Turns out we were good at convincing ourselves we were good at this. The gap between reported CAC and real CAC was embarrassing." - Head of Growth