Plan an experiment before launch, record results, and make a decision against agreed criteria. Use before exposing a product change to a defined group of users.
OverviewIdentify the experiment owner and planned dates.
HypothesisState the change, intended audience, expected outcome, and the measure that will show whether it happened.
VariantsDescribe what each group will experience and how eligible traffic will be split.
Primary metricChoose one outcome measure that determines whether the experiment succeeds.
Guardrail metricsList measures that must stay within acceptable bounds, even if the primary metric improves.
Audience and allocationDefine who can enter the experiment, how assignment works, and what share of eligible traffic will be included.
Sample size and durationRecord the effect size the experiment can detect, the planned sample, and expected runtime.
Decision ruleWrite the outcomes that lead to shipping, iterating, or stopping.
ResultsComplete this section after the planned end date, or earlier if a guardrail threshold is breached.
DecisionRecord the decision and the next actions.