Card disputes: trace reversals, evidence and final recoveries
A merchant can report strong dispute recoveries while cash remains tied up and many cases are unresolved. The useful research object is a dated cohort of disputed amounts, linked back to the original transactions and forward to their final disposition. Monthly recoveries alone do not establish a recovery rate or a reduction in the final economic loss.
This guide builds an illustrative merchant ledger. All amounts, dates and outcomes in the example are invented currency units, with no exchange-rate effects. It uses Visa merchant guidance as a process reference, not as a universal description of every network, jurisdiction or merchant contract.
Locate the process and version
The Visa dispute-resolution page explains the merchant’s role in responding through the acquiring relationship. The Dispute Management Guide for Merchants, October 2026, locates the simplified life cycle on printed page 6, response guidance on pages 7–8, and the relevance of condition and timing on page 12.
The guide’s legal notices state that the Visa Rules and merchant agreement take precedence. Therefore record the actual dispute condition, processing stage, acquirer instructions and applicable version before interpreting a merchant’s case status. This worksheet does not derive a response deadline from a generic dispute label or promise that submitting evidence will win.
Rebuild a complete sales cohort
Assume January completed sales of 1,000. In February, disputes generate debits of thirty against identified January transactions. In March, twelve is finally recovered after the relevant process is complete, eighteen is finally lost, and non-refundable dispute fees total three. Assume no other refunds, reserves, taxes or adjustments in this isolated example.
The cash bridge attributable to the cohort is 1,000 − 30 + 12 − 3 = 979. The final principal loss is eighteen, and the additional fees are three, giving twenty-one of combined economic leakage. Twelve recovered is not an extra sale; it reverses part of an earlier debit. Recognising both a new twelve of revenue and a reduced dispute loss would count the same recovery twice.
| Cohort movement | Cash effect | Interpretation |
|---|---|---|
| January sales settled | +1,000 | Original transaction cohort |
| February dispute debits | −30 | Amount challenged and debited |
| March final recovery | +12 | Part of thirty restored |
| Non-refundable fees | −3 | Separate expense assumption |
| Net cohort cash | 979 | Under the isolated assumptions |
For these fully resolved amounts, principal recovery is 12 ÷ 30 = 40%. Principal loss relative to original sales is 18 ÷ 1,000 = 1.8%; including fees gives 2.1%. Each ratio has its own denominator. The loss ratio does not include the merchant’s cost of goods, time spent responding or lost future business, which remain outside the supplied inputs.
Keep three clocks and two cohorts
Record original transaction date, dispute opening and final outcome separately. The same January sale can affect cash in February and March. A cash report grouped by March receipt date therefore answers a different question from a January sales-cohort loss report or a February dispute-opening performance report.
For a separate February dispute cohort, suppose twenty was debited, eight has finally been recovered, four finally lost and eight remains unresolved. The amount-based reconciliation is 8 + 4 + 8 = 20. Recovery among resolved principal is 8 ÷ 12 = 66.67%, while recovery so far against all disputed principal is 8 ÷ 20 = 40%. Neither is an estimate that the unresolved eight will recover at the same rate.
Keep cohort age alongside each comparison. An older cohort may have more losses recognised simply because more cases have finished. Comparing it with a young cohort and calling the difference operational improvement would confuse maturity with outcome. Use common observation windows where feasible, while still displaying the unresolved balance explicitly.
Test provisional credit and reopening
Suppose the twelve in the main case was provisionally credited in March but could still be reversed. Cash has increased, but final recovery remains unconfirmed. Maintain separate credited-cash and final-disposition columns. If the twelve is subsequently reversed, link the debit to that existing case; do not create another thirty of disputed sales.
The case can also contain partial outcomes. If a transaction for ten has six disputed and only four recovered, the sales amount is ten, disputed principal six and final principal loss two. Counting it as one won case would obscure the partial loss; counting it as one lost case would obscure the recovery. Report amount and case-count metrics separately with explicit treatment of partial cases.
A closed flag in an internal queue may mean the merchant’s response work is finished rather than the financial process is final. Obtain the acquirer outcome and settlement reference supporting the status. Where they are unavailable, use a qualified status instead of treating administrative closure as irreversible cash recovery.
Match refunds and fees without duplication
For a different boundary, assume a customer has already received a documented refund of six, and the same six is later disputed. Do not add the refund and dispute automatically as two final losses. Match the original transaction, refund reference, dispute and final credit or reversal to establish which cash movements remain outstanding.
A refund initiated but not settled is different from a completed refund. Likewise, a fee may be assessed per case, per stage or under a separate contract, and may not be refunded when principal is recovered. Use actual fee entries and contractual evidence. The three-unit fee in the main case is an explicit example input, not a Visa-wide fee schedule.
If multiple currencies occur, keep principal and fees in their original currencies and record the settlement conversion separately. An exchange difference is not a successful dispute recovery. The isolated numerical bridge intentionally avoids that complication so every movement can be traced to one disputed principal balance.
Use the dispute verification sheet
| Field | Obtain | Reconciliation action |
|---|---|---|
| Transaction | Original sale and settlement ID | Link disputed portion to original amount |
| Case | Network/acquirer reference and condition | Track stages without duplicating principal |
| Evidence | Submitted materials and acknowledgement | Submitted distinct from accepted or final |
| Cash | Debits, credits and reversals | Separate provisional and final movements |
| Outcome | Amount-based final disposition | Recovered, lost and unresolved sum to cohort |
| Refunds and fees | References and contract terms | Avoid duplicate refunds; retain actual costs |
Start from a defined transaction population, obtain the acquirer’s case export and join by stable identifiers. Investigate unmatched cases before calculating rates. A missing original transaction cannot safely enter a sales-cohort denominator, while a missing final outcome belongs in the unresolved balance rather than a presumed win or loss.
When dashboards and settlement reports disagree, compare gross versus net amounts, report cutoff, currency, partial cases, provisional credits and fee inclusion. Save the unreconciled amount and the record required to resolve it. A dashboard percentage cannot substitute for a cash movement that has not appeared in the settlement evidence.
Update the ledger after new disputes, accepted documents, actual settlement movements, final decisions or reopened cases. Preserve the prior observation date and cohort age. The resulting conclusion should identify cash tied up, final principal loss, fees and unresolved exposure separately; it should not turn a recent recovery inflow into an unsupported claim about all future disputes.
Define the amount population before measuring performance
The amount ledger should carry a controlled case identifier and a separate original-transaction identifier. Two disputes concerning different portions of the same sale can be separate cases without creating two original sales. Conversely, two status exports for the same case are observations of one process. Keep both dimensions so a deduplicated transaction count does not erase legitimate partial-case detail.
For the unresolved-eight example, set an explicit follow-up observation. If five is later finally recovered and three finally lost, the cohort ends with thirteen recovered and seven lost, totaling twenty. Final principal recovery is then 65%, replacing the earlier recovery-to-date measure of 40%. The change comes from actual outcomes, not from deleting the unresolved amount from the denominator.
Preserve the earlier twelve of resolved principal and its 66.67% resolved-only rate as a historical observation, with its name. Do not compare that selected-subset rate to the final all-principal 65% and conclude performance deteriorated. They answer different questions. A useful report states which observation window and amount population support every percentage, alongside the actual cash movements.
FAQ
Does a provisional credit count as a final recovery?
Only cash movement is established until the remaining outcome uncertainty is resolved and documented.
Can recoveries this month be divided by this month’s new disputes?
That mixes cohorts unless the amounts concern the same defined population and observation window.