How to detect duplicate supplier invoices

The same invoice received by email AND by mail. A vendor who resends an invoice because they think they were not paid. A mistyped number that makes an invoice look brand new. Duplicate supplier invoices almost never come from fraud: they come from manual processes and small human errors. And once paid twice, they are slow to claw back. Here is how to catch them before they cost you.

Why duplicates happen

The culprit is not carelessness, it is manual processing. Research on human error confirms it: the moment a person keys or matches invoices by hand, an error rate sets in, between 1% and 5% in Panko's work, and around 1% for single entry (Barchard and Pace, 2011). Add several intake channels (email, mail, portal), several people handling the same vendors, and resent invoices, and duplicates become inevitable.

An honesty note about the numbers. You will read everywhere that duplicates represent "X% of spend." In reality, those estimates vary wildly from one source to the next (from 0.1% to several percent) and rarely rest on a solid primary study. We prefer not to put forward a figure we cannot verify. What is certain is that the problem is real, well documented, and entirely avoidable.

How to detect duplicates, in practice

Spotting a duplicate by eye, across hundreds of invoices a month, is impossible. Software does it continuously, on several criteria:

1. Matching on key fields. Same vendor, same invoice number, same amount, same date: the software flags the repetition instantly.
2. Cross-channel detection. An invoice received by email and another by mail are compared, even if they arrive at different times.
3. Fuzzy matching. A number with a typo, an amount off by a cent: the software flags the resemblance instead of missing it.
4. Alert before payment. The duplicate is blocked and flagged during processing, not discovered months later in an audit.

Detect, but above all prevent

The best duplicate detection is the one that keeps the duplicate from existing. By centralizing invoice intake in a single platform, reading every invoice automatically (OCR extracts, AI contextualizes) and matching each one to the purchase order, you remove the very conditions that create duplicates. People only validate the flagged, doubtful cases, with an audit trail.

The trickiest duplicates

Obvious duplicates, the same invoice received twice, are the easiest to catch. The real traps are elsewhere: an invoice resent with a slightly changed number, an invoice and its statement showing the same amounts, a credit note matched to the wrong bill, or the same expense surfacing in two entities of the same group. To the eye, these almost always slip through. Software that compares amounts, dates and vendors with fuzzy matching flags them, even when nothing is strictly identical. It is this grey zone, invisible in manual processing, that costs the most over time.

In short

Duplicate supplier invoices are a process problem, not a people problem. You catch them by automatically comparing key fields across every channel, and you block them before payment. Better yet, you prevent them by centralizing and automating processing. That is what specialized, proprietary accounts payable software, built and hosted in Canada, is for.

Book a free meeting

Frequently asked questions

Sources

• Barchard, K. A., & Pace, L. A. (2011). Preventing human error: The impact of data entry methods on data accuracy and statistical results. Computers in Human Behavior, 27(5), 1834-1839. Abstract (DOI 10.1016/j.chb.2011.04.004)
• Panko, R. R. Research on human error and data entry error rates. arXiv:0801.3114