A duplicate invoice bills for something that has already been invoiced. The danger is specific: unlike most exceptions, which stall an invoice, a duplicate that clears results in the company paying twice. Duplicate detection is the check that catches them before payment.
Most duplicates are honest mistakes, and the ones that do the most damage are the ones that do not look identical.
A duplicate rarely looks identical
The clearest case is the same invoice number from the same supplier submitted twice. Harder cases are near-duplicates: the same charge sent under a slightly different invoice number, a resubmission with a new date, an invoice sent once by email and once by mail, or the same work billed on both a final invoice and an earlier progress invoice. All of these bill for the same underlying spend.
High volume makes duplicates inevitable
A supplier who does not get paid quickly assumes the first invoice was lost and sends another. An invoice arrives through two channels and is entered twice. A credit and a rebill get tangled. A buyer and an AP clerk both key in the same document. None of these requires bad intent, and at high volume they are close to inevitable.
Exact-match detection catches only the obvious ones
Most systems check for duplicates by matching on a few fields, usually supplier, invoice number, and amount. That catches the identical case and misses the rest. Change the invoice number by one character, split the amount across two lines, or use a different supplier record for the same vendor, and the exact-match test passes the duplicate straight through. Catching near-duplicates requires comparing the substance of the invoice against history, rather than a few key fields.
A cleared duplicate is money out the door
A duplicate that clears is a straight loss until someone notices, and recovery after the fact is slow and often incomplete. Companies run duplicate-payment audits and hire recovery firms specifically to claw back money that should never have left.
Real detection compares every invoice against history
Prevention starts with clean vendor master data, so one supplier is not stored under several records, and with a single intake channel so the same invoice does not enter twice. Detection then has to go beyond exact matching to compare each new invoice against the full payment history for that supplier, including the near-duplicate patterns a field-level check ignores.
Fragment catches duplicate invoices autonomously, including the near-duplicates that exact-match rules miss, by comparing each invoice against the full payment history. See how it works or request a demo.