A/B testing setup
📖 3 min readUpdated 2026-04-19
A/B testing splits traffic between control (A) and variant (B), measures outcomes, picks winner.
Prerequisites
- Clear hypothesis
- Single variable changed
- Minimum sample size calculated
- Primary metric defined
Setup
- 50/50 traffic split
- Random assignment
- Both variants tracked identically
- Run until statistical significance
Post-launch
Don't peek early, wait for statistical significance before declaring winner.