Kinda nerdy but—ever heard of the "Simpson’s Paradox"? Aggregated data can show one trend while the subgroups tell the opposite story. Wild, right?
Happened to me with A/B test results. Now I never trust the overall metric without checking the segments.
Hot tip: Amplitude’s segmentation tools are clutch for this.
Happened to me with A/B test results. Now I never trust the overall metric without checking the segments.
Hot tip: Amplitude’s segmentation tools are clutch for this.
