Choose a cohort and a close date
Group cases by opened date or another agreed cohort rule, and use the same rule across periods. Record the status as of a fixed date. Keep pending cases visible and label comparisons provisional until the cohorts have had comparable time to mature.
Deduplicate by processor and case ID, then link alerts or inquiries that refer to the same payment without treating them as separate chargeback wins. Track account, currency, reason and disputed amount so finance can segment and reconcile the data.
Separate win rate from coverage
One useful win rate is won cases divided by decided contested cases. Coverage asks what share of eligible cases you submitted. Report both. A team can show a higher win rate after submitting fewer difficult cases, while recovering less overall.
Illustrative example: 100 cases are eligible. You contest 80. Of those, 40 are won, 30 are lost and 10 are pending. Submission coverage is 80%, and the decided contested win rate is 40 ÷ 70, or 57.1%. Forty wins represent 40% of the eligible cohort as of this snapshot. No merchant results are implied.
Reconcile dollars by outcome type
Show chargeback principal returned separately from funds kept after qualifying inquiries. An alert resolved by refund belongs in the prevention ledger. Subtract applicable service fees and processor costs when calculating net recovery, and document the treatment of fee refunds and currency conversions.
Illustrative example: $1,000 returned after a chargeback win yields an $800 amount after a 20% Safe fee and before processor fees or other costs. A qualifying $1,000 inquiry has the same fee at the standard rate, but the retained amount belongs in a separate column because the principal was never debited.
Make the weekly review actionable
Review coverage, approaching deadlines and missing-evidence reasons weekly. Compare closed cohorts monthly by provider, store and reason. Pair every percentage with its case count, and use a consistent exchange-rate rule before aggregating multiple currencies.