ds-requesting-analysis-review

An independent analysis review verifies the reproducibility of experiment design, statistical methods, metric definitions, and interpretation before final decision readiness. It checks randomization alignment, invariant signals, and sensitivity to outliers to prevent flawed analytics from hardening into business calls.

To request an analysis review, provide a complete review package containing your question context, experiment unit, metric formulas, data windows, exclusions, validation level, and notebook or SQL artifact references. This allows an independent reviewer to verify your statistical methods end to end.

A metric validation review package requires your study question, decision context, hypothesis, experiment unit, final metric formulas, data windows, exclusions, validation level, and exported output tables. These artifacts enable thorough checks for unit alignment, randomization correctness, and statistical method suitability.

Yes, methodology proposals require the same reproducibility verification as experiment analyses. You can request a decision-stage review to check statistical method correctness, sensitivity to tails, and invariant signals before your recommendations are published and finalized.

Sample Ratio Mismatch (SRM) signals and invariant checks must be verified during an independent methodological review before decision readiness. Checking SRM ensures your randomization alignment is correct and your experiment design supports valid, reproducible statistical conclusions.