The Creative Testing Budget Planner
Testing twelve concepts on a $6,000 budget does not give you twelve answers. It gives you twelve guesses. This works out how many you can actually afford to read.
The most common creative mistake is not making too little — it is spreading the test budget so thin that no concept ever accumulates enough conversions to tell you anything. You end up killing winners and scaling noise.
A “concept” is a distinct angle or idea, not a variation. Three hooks on the same script is one concept tested three ways, and they should share a budget.
The 20-conversion rule
Below roughly 20 conversions, the confidence interval around a concept's CPA is so wide that it overlaps with almost every other concept in the test. You are not measuring creative quality, you are measuring variance. At 20 you can separate the clear losers; at 50 you can start ranking the middle.
This is why the honest answer to “how many creatives should we test?” is always “how much are you spending?” A brand at $8,000 a month cannot run the same testing programme as a brand at $150,000, no matter how good the creative team is.
Test fewer, bigger ideas. Four genuinely different angles with enough budget to read beats twelve variations that all die inconclusive. Save the variation testing for after you have found an angle that works.
How we structure a month
| Week | What happens |
|---|---|
| Week 1 | New concepts go live in the testing campaign at even budgets. Nothing is judged. |
| Week 2 | Clear losers are cut once they pass the conversion threshold. Survivors keep running. |
| Week 3 | Winners graduate into the scaling campaign. Variations of winners get briefed. |
| Week 4 | Readout: what worked, what the angle was, what to make next. This feeds the next month's briefs. |
Three rules that save money
- Never judge on day one. Meta's delivery is unstable in the first 24–48 hours. A concept that looks dead on Tuesday is frequently mid-pack by Friday.
- Test angles, not executions. Changing the hook, the claim or the audience the ad speaks to produces real signal. Changing the font does not.
- Write down the hypothesis before it launches. If you cannot say what you expect to learn, you are producing content, not testing.
Creative volume is the ceiling on Meta performance, which is why Ads + Creative exists as a package. We brief, produce and read 30–60 assets a month against a documented angle map.
Creative Is The Bottleneck
Tell us your spend and we will show you the testing programme it can actually support.