In an A/B test, half of the visitors see version A and the other half version B. They differ in one thing - a headline, a button, a price, an image. After enough visits you compare which version brought more orders or leads.
When it makes sense
A test needs traffic. With a few dozen visits a week the result cannot be told apart from chance, and the test only delays the decision. Then it is better to make the change from experience and watch the trend.
The most common mistake
Stopping the test the moment one version briefly leads. The gap often disappears within a week. Decide the test length and the sample size up front.
