You are reviewing an experiment on a rider growth surface, such as the Didi Chuxing new-user coupon flow. The team sees a metric move and wants to know whether it is real and whether it matters enough to ship.
How would you decide whether a metric change is statistically significant and practically meaningful?
Hypothesis testing for an experiment metricInterpreting p-values and confidence intervalsSeparating statistical significance from business impactUsing a pre-defined practical threshold such as MDE