Experiment Duration

Experiment duration is the span an experiment must run to gather enough data for a reliable result — neither shorter nor longer.

Experiment duration is a statistical decision, not a feeling: it is computed from the required sample size and daily traffic, and cutting it short risks a false result. Two symmetric errors live here: stopping early just when the number looks good (which recasts chance as certainty), and dragging the experiment on endlessly until it finally turns 'significant'. Beyond volume, the duration should cover at least one full weekly cycle, because weekend behavior differs from midweek; an experiment run for only three weekdays misses the weekend pattern. The rule is to set the duration before starting and hold to it.