The creative testing system behind 4× ROAS accounts

The problem with "more ads"

Most accounts don't have a budget problem. They have a creative problem. When ROAS drops, the usual fix is to launch more variations of the same idea, which spends money to learn very little. The platforms now handle targeting well. What they can't do is invent a better reason to buy.

So we treat creative as the main lever and run it like a lab: one clear question per test, a fixed budget, and a decision date set before anything launches.

20+ new ads / month · 72h to a decision · 4.1× median ROAS

How we structure a test

Every test changes one thing. If the hook, the offer and the format all change at once, a winner tells you nothing you can use again.

The three layers

• Angle: the reason to buy, such as price, proof, identity or problem.

• Hook: the first three seconds, tested in batches of five.

• Format: UGC, founder‑led, static or demo. We change this last.

We test angles first because they move results the most. Hooks come next, and format is the final polish.

Budget rules that stop the bleeding

A test campaign gets 15 to 20 percent of spend. Each ad gets enough budget to reach about 50 link clicks before we judge it. If it's below half the account's target ROAS by then, it's off. If it beats the target, it moves to the scaling campaign the same day.

"Kill losers fast and promote winners faster. Most accounts do neither, so they pay for the same lesson every month." — James Dunn, Founder

Reading the results

Platform ROAS is a signal, not the truth. We check every winner against blended numbers: total revenue divided by total spend, plus new‑customer rate from Shopify. An ad that only retargets people who were going to buy anyway isn't a winner.

What we report each Monday

• Blended ROAS and contribution margin

• Top three ads by profit, not by clicks

• What we're testing next, and why

Putting it to work

Start small. Pick one angle you've never tested, brief five hooks for it and give them a week. Then write down what you learned before you spend another dollar.

Key takeaway

Treat creative as the targeting. Test one variable at a time, judge on blended profit, and move winners into scale within 24 hours.