A/B Testing

A/B testing — also called split testing — is the practice of showing two versions of a page, element, or flow to different segments of your audience simultaneously, then measuring which version drives more conversions. It's the closest thing to a controlled experiment that most businesses can run on their own website.

Done correctly, A/B testing removes guesswork from conversion optimisation. Instead of debating whether the headline should say "Start your free trial" or "Get started free," you test both and let your actual visitors decide. The version that converts more wins — not the one the founder preferred, not the one the designer liked, but the one that works.

The challenge is that most teams run A/B tests incorrectly. They test too early (before reaching statistical significance), stop tests too early (the peeking problem), or test cosmetic changes that never move the needle. A button colour change rarely has a meaningful impact on revenue. A headline rewrite, a restructured pricing page, or a simplified checkout flow often does.

The articles in this category cover the mechanics of rigorous A/B testing: how long to run a test, what sample sizes you actually need, which mistakes silently invalidate most tests, and which tools work best for different platforms and team sizes. If you're running tests and not seeing consistent winners, start here.