What is A/B Testing?
A/B testing is a controlled experiment that compares two versions of a product experience by randomly splitting users between them and measuring which performs better against a defined metric. It isolates the effect of a single change so teams can decide from evidence rather than opinion.
Reviewed by Gensudo Team · 23 July 2026
In more depth
In an A/B test, users are randomly assigned to a control (A) and one or more variants (B), and a target metric is compared, using statistical significance to distinguish real effects from noise. Reliable testing requires a sufficient sample size, a pre-declared metric and guardrails to catch harm elsewhere. Randomisation is what lets teams attribute any difference in outcomes to the change itself.
Why it matters
Intuition about what will improve a product is frequently wrong, even among experts, and shipped changes can quietly reduce key metrics. A/B testing replaces debate with measurement, protects against confidently shipping harmful changes, and, at scale, turns a product into a learning system where small validated improvements compound over time.
A product example
A team tests a shorter checkout headline against the original on half of traffic. The variant lifts completed purchases by a statistically significant 4%, so it ships, a change that a design review alone might have dismissed as trivial.
Documents where this shows up
Validation Findings Report for Product Managers · Product-Market Fit Assessment for Product Managers
Related terms
Product Validation · Cohort Analysis · KPI (Key Performance Indicator) · Activation Rate