Ad glossary
A/B testing
Comparing two versions of an ad, page or element by splitting traffic between them to see which performs better on a chosen metric.
A/B testing, also called split testing, compares two versions of something by showing each version to a separate, comparable group and measuring which performs better. The versions might be two ads, two headlines, two landing pages or two offers.
How a good test works
- Pick one question, for example whether a review-led hook beats a problem-led hook.
- Change one main variable so you know what caused the difference.
- Choose a primary metric in advance, such as CPA, CTR or conversion rate.
- Split the audience so each person sees only one version where possible.
- Run until you have enough data, rather than stopping at the first sign of a winner.
- Record the result and what you will do with it.
A/B testing on Meta
At the time of writing, Meta offers A/B testing through Ads Manager and its Experiments tools, which split audiences so the same person is not shown both versions. Simply running two ads in the same ad set is not a true A/B test, because the delivery system pushes spend towards whichever ad it predicts will do better, and the groups are not randomised.
Common pitfalls
- Declaring a winner from a small number of conversions.
- Testing tiny variations when bigger ideas would teach more.
- Changing several things at once and not knowing which mattered.
- Ignoring outside factors such as a sale, a holiday or a stockout during the test.
A/B testing is one part of a wider creative testing process and is closely related to incrementality testing.
Last reviewed Oct 27, 2026