A/B Testing

Definition

A/B testing is a controlled experiment where visitors are randomly split between two versions of a page (A and B) that differ by one element, and you measure which version produces more conversions. It replaces guesswork with evidence about what actually works on real users.

In depth

In an A/B test, half your traffic sees the original (the 'control') and half sees a variant with one deliberate change — a different headline, button, image, or form length. Because the split is random and runs at the same time, outside factors like seasonality or ad changes affect both versions equally, so any difference in conversion rate can be attributed to the change you made. You run it until you have enough data to trust the result.

It matters because opinions about what 'should' convert are often wrong, and the cost of guessing is real money. A/B testing lets you make changes you can defend with data instead of hunches, and it builds a record of what your specific audience responds to. Over time, a series of small validated wins compounds into a much stronger page.

The biggest mistake is calling a test too early. If you stop the moment one version is ahead, you're often just reacting to random noise — you need enough conversions for the result to be statistically meaningful. WellBuilt also tests one change at a time so we actually know what caused the lift; changing five things at once tells you the page got better but not why.

Worked example

Related terms