Split test (A/B test)

A split test, or A/B test, shows two versions of a page or ad to people at random and compares the results. Optimizely describes it as showing two variants at random, then using statistics to judge which does better against a goal.

The limit is traffic. With few clicks, a gap between versions can be chance, so a test needs enough visitors before anyone names a winner.

Example: a tour operator runs two advertorials for the same safari, one a story and one a list. Each gets half the clicks, and the one that brings more inquiries wins. On Itinex, two page versions run on an even split until the numbers name a winner (the platform).

Related: landing page · click-through rate · conversion rate