Variance Reduction for Heavy-Tailed Monetization Metrics in Ranking Experiments via Post-Stratification

In three linesVariance reduction method for heavy-tailed monetization metrics in ranking experiments. Combines post-stratification with CUPED on pre-experiment covariates. Deployed at ShareChat: reduces variance by ~45% without additional traffic.

Summary generated by Claude — human-verified