Good checks start with the right grain of data: a settled payment allocated to a specific invoice, or one remaining balance on a credit memo.

Payment file

Use a payment-allocation report from your accounting system. An invoice register alone does not prove payment. A bank export alone usually lacks the invoice reference. Combine or export these records in your accounting workflow before importing.

ColumnWhat it means
entityThe legal entity that paid. Keep distinct clients separate.
vendorA consistent supplier name. Different spellings are not merged automatically.
invoice_referenceThe invoice the allocation settled.
payment_idThe actual transaction reference. Essential for distinguishing two payments from two export rows.
datePayment date, YYYY-MM-DD.
amountThe settled allocation, in major units such as 1234.56.
currencyUSD, EUR, GBP, CAD, AUD or NZD. No automatic conversion.
statuspaid, settled, cleared, completed, reversal, or an explicit void/unpaid status.
invoice_totalOptional original full invoice obligation, including tax and adjustments. A partial amount is not a valid total.

Supplier credits and your ledger

Both files use entity, vendor, credit_reference, date, open_amount and currency. Use a positive remaining balance in open_amount. Do not upload several monthly snapshots of the same credit. A fully applied credit has no open balance.

Before running the scan

  1. Use the same cutoff date and consistent entity/vendor names.
  2. Remove subtotals and report footers.
  3. Export UTF-8 comma-separated CSV. Use dot decimals and ISO dates.
  4. Represent reversals as negative amounts with status reversal.
  5. Include all allocations within the invoice scope, so an apparent overpayment is not a partial picture.

When your source uses different column names, map the columns in the workspace. CreditLatch does not guess missing payment evidence or silently skip malformed amounts.