Mockingbird

Learning Center

How creative testing works on Meta

Creative testing is how advertisers find out which ads work before they put more money behind them. On Meta there are two broad methods. You can run several ads in one ad set and let the delivery system choose, or you can use a controlled test, either Meta's A/B test or its newer creative test, which gives each version a fair share of people and budget.

This topic explains how each method splits people and budget, what its result can and cannot tell you, and how to judge whether a gap between two ads is real or noise. It covers the mechanism. The guides listed below cover the steps and the budget.

Key terms

Common questions

What is creative testing in Meta ads?
Creative testing is running different ad versions to learn which one produces better results, usually a lower cost per purchase or lead. It can be as simple as several ads in one ad set or as controlled as Meta's A/B test. The aim is to find ads worth scaling and to learn what to make next.
How many ads should I test at once on Meta?
Meta's A/B test and its creative test both allow up to five versions, and three to five is a common working range. Each extra version splits the budget further, so every version takes longer to gather enough purchases to compare. Small budgets are better spent on fewer, bigger differences.
What metric should decide a creative test?
Use the result you pay for, such as cost per purchase or cost per lead, which is also what Meta's A/B test uses to pick a winner. Click-through rate and other early metrics help explain why an ad won. They should not decide the test on their own.
What is a winning ad?
A winning ad beats the alternatives on the result you care about, with enough data that the win is unlikely to be chance. On Meta that usually means a lower cost per purchase or lead over a test long enough to trust. Winners fade as frequency rises, so the next test should be ready before they do.

Put it to work

Other topics