Definition
A controlled experiment in which two versions (A and B) of an element (page, flow, button, copy, price) are compared by presenting them randomly to different users to measure which performs better on a target metric. It is the canonical empirical method for making optimization decisions based on statistical evidence rather than intuition. Variants: multivariate tests (several variables at once), multi-armed bandits (dynamic allocation).
Relationships
types-of / more specific
—
same semantic field
—
Concepts linked to this one by a real relationship in the glossary. Hover over each one to see why it's related.