The Average Hides the Reversal
Detect an aggregate metric that should be segmented before decision-making.
The overall conversion winner may be hiding a device-mix reversal. Find the dashboard region that should trigger a Simpson's paradox check. The large top-line conversion result Correct. A single aggregate result needs segment rates and weights before rollout. The rollout notes on the table The rollout notes are a consequence. The evidence problem is the aggregate conversion rate. The marketer presenting the result The presenter is not the issue. The metric structure is. The smaller segmented panels Those panels are part of the fix. The problem is trusting the top-line number without reading them. aggregate skepticism Practiced spotting when an…
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in