When a fintech issues cards (e.g., a spend management platform), it sits in the flow of funds. It must reconcile three data sets: The Processor (Marqeta/Lithic), the Network/Scheme (Visa/Mastercard), and the Sponsor Bank. The complexity lies in the fees. The amount the user spends ($100) is not the amount that settles ($98.50).
The Settlement File (TC33 / IPM)
Visa/Mastercard send massive daily files (TC33 or IPM) detailing the net settlement.
Interchange Income: The merchant pays this. It is revenue for the fintech.
Scheme Fees: Visa charges this for using the network.
The Net Settlement: The bank debits the fintech's account for Spend - Interchange + Scheme Fees.
Recon Logic: The engine must parse the raw Scheme file to verify that the Processor (who creates the API transaction) accurately calculated the Interchange. A 0.1% variance here equates to massive revenue leakage.
Clearing vs. Authorization Matching
Auth: User swipes for $100. Ledger creates Pending hold.
Clear: 2 days later, merchant settles for $120 (added tip).
The Match: The reconciliation engine must match the $120 clearing record to the $100 auth based on Approval_Code and Card_Token, then update the user's ledger for the extra $20.
