Explores how A/B test results often fail to replicate due to novelty effects, sampling bias, and misaligned success metrics. Bridge this with real-world CRO experience and behavioral noise.