You work on a digital product team that runs frequent A/B tests and often sees conflicting reads from the same experiment depending on when the results are checked. Some launches look strong at first, then flatten out, while others show wins in the primary metric but raise concerns in related metrics.