A/B test (also called split testing) is an experiment that shows two versions of something — A and B — to different visitors and measures which one performs better against a goal, such as clicks, sign-ups, or purchases.
You change one main variable at a time: a CTA label, headline, form length, or button color. Traffic is split so results are comparable, then you keep the winner (or keep testing). It’s a practical way to improve conversion and UX with evidence instead of opinions.
A real-life example of an A/B test
A bakery’s site shows half of visitors “Order now” and the other half “Reserve for pickup.” After enough visits, version B gets more completed orders. That isn’t a vibe check — it’s measured behavior. Large companies run thousands of these tests; small businesses can start with one clear change and a solid traffic sample.
A/B tests work best with a defined metric, enough volume, and patience — ending too early can mislead. Heatmaps and analytics help you decide what to test. For more on testing culture, see Culture of Experimentation.