PI-003 ยท Distribution and supply chain
Debit disputes, chargebacks and provisional credit
A tide that only ever went one way.
Presenting problem
We reconstruct debits, chargebacks, provisional credits, payments, adjustments, and contractual calculations to determine whether the amounts being asserted are supported by the underlying records.
What brought us in
Dispute losses exceeded forecast by a wide margin with no corresponding change in dispute volume.
What the client believed
That fraud pressure had increased and that the loss rate reflected external conditions.
What the evidence showed
Dispute volume and external fraud pressure were flat. The excess came from disputes that entered the process twice through separate intake paths, and from provisional credits that were never reversed after the dispute was decided in the institution's favour.
How the examination proceeded
Method applied to this engagement.
- 01Preserved dispute intake, network chargeback, provisional credit, reversal and settlement records.
- 02Matched dispute events across intake channels on transaction identity rather than case identity.
- 03Traced each provisional credit to its decision and to its reversal, or to the absence of one.
- 04Tested the duplicate population for legitimate re-presentment before including any item.
- 05Quantified unresolved credits and duplicate credits separately, with no overlap between the two totals.
Substantiated findings
What the records support.
Provisional credits remained outstanding after the underlying dispute was decided against the cardholder.
Substantiated - documentaryA population of disputes was credited through two intake channels for the same underlying transaction.
Substantiated - duplicateNetwork representment rights lapsed unused on a measurable subset of eligible disputes.
Substantiated - contractualExposure
$32m+ disputed exposure, $8m+ in credits challenged
Figures are rounded and generalized for publication. In delivery, each figure is deterministic and carries its inputs, calculation version and population definition.
Control failure map
Why it was possible.
- Two intake channels shared no common key.
- Provisional credit reversal was a manual step with no exception report.
- Representment deadlines were tracked per case, never in aggregate.
Outcome
What the client did with it.
- Loss forecast rebuilt on mechanism rather than volume.
- Duplicate population recovered directly.
- Reversal step converted into a controlled exception queue.
Every duplicate pair named in the report resolves to two preserved source records and the calculation version that joined them.
