Replaced hero-ad roulette with a creative engine: 300+ creatives tested in six months on a strict one-variable cadence.
ROAS held while ad spend tripled
A supplements brand that had found a few great ads and built a business on top of them. The problem with winning creatives is that they expire — and nobody knew how to make the next one on purpose. Scaling meant spending more on ads that were already tired.
- 01The whole account riding on three creatives, all past their peak.
- 02No angle bank — every new ad was a guess, not a hypothesis.
- 03Tests that changed the hook, the visual and the offer at once, so nobody ever knew why anything won.
- 04Creative production that started from a blank page every single time.
Bucket first. Then the tap.
Mined reviews, support tickets and social comments into an angle bank — every claim, fear and desire the customer ever voiced, ranked.
Built the production pipeline: UGC briefs, static templates and hooks generated from the angle bank, not from moodboards.
Installed a weekly testing cadence — one variable per test, every result logged against its hypothesis.
Promoted winners into scaling campaigns on a schedule, and retired fatigue before it showed up in the P&L.
The insight, not the settings screen.
Winning ads aren't found, they're manufactured. When creative runs as a system — angle bank in, tested learnings out — fatigue stops being an emergency. It's just Tuesday.