Split Test (A/B Test)

A split test, also called an A/B test, is a way to compare two versions of something by showing each to a different set of visitors at the same time, then seeing which one performs better. Instead of arguing about which headline is stronger, you let real behavior decide.

Half of visitors  ->  Version A (green button)
Half of visitors  ->  Version B (blue button)
Compare which version got more signups

You can split test almost anything that affects results: a headline, a button color, an email subject line, the price shown, or the whole layout of a landing page. The rule that keeps it honest is to change one thing at a time. If you change five things at once and version B wins, you have no idea which change caused it.

The other rule is patience. A split test only means something after enough people have seen both versions. If version A got three signups and version B got five, that could easily be random luck, not a real difference. You need a decent number of visitors before you trust the winner.

Split testing is how good marketers improve over time. Small, tested improvements to a page's conversion rate compound, and they beat guessing. To understand the number a split test is trying to move, read conversion rate explained.