Process Deep Dive

Catching Duplicate Payments Before the Money Leaves

A duplicate payment is easy to make and hard to claw back. The fix is not another approval step but catching the duplicate before the run, across every channel it could have entered.

An agent catching a duplicate payment before the payment run
Easy to send
Hard to recover
the asymmetry
Many channels
Where dupes enter
not one queue
Before the run
Where to catch it
not after
Fuzzy match
Not exact
the real dupes
The short version
  • A duplicate payment leaves easily and comes back slowly, if at all. Recovering it means a vendor's goodwill and weeks of chasing, so prevention is worth far more than detection after the fact.
  • Duplicates are hard to catch because they rarely look identical: the same invoice enters through email and a portal, or with a slightly different invoice number, vendor spelling, or amount.
  • The fix is a check that runs across the full payable population before the payment run, matching fuzzily on vendor, amount, date, and invoice reference rather than only on an exact key.
  • Agents run that check continuously and hold suspected duplicates for review before the money leaves, with a person deciding the genuine edge cases.

Duplicate payments are the quiet leak in accounts payable. One invoice comes in by email and again through a vendor portal, or a bill gets entered twice under slightly different details, and both get paid. The money leaves in seconds. Getting it back takes a vendor willing to return it and weeks of your team's time chasing a credit.

The usual response is another approval layer, which slows everyone down and still misses the duplicates that do not look identical. The better fix is a check that runs across everything payable before the run and catches the near-duplicates, the ones with a different invoice number or a transposed amount, that exact matching never finds.

Duplicate PaymentsAccounts PayableControlsFraudPayment RunReconciliation
01

Why duplicates are so easy to make

Many front doors

AP has many front doors. Invoices arrive by email, through portals, as PDFs, and sometimes on paper, and the same invoice can enter through more than one. A vendor resends because they did not see the first payment; a team member enters a bill that was already captured; a credit and a re-bill get tangled. None of it requires anyone to be careless.

Because the entry points are separate, a duplicate entered through a different door does not collide with the original in an obvious way, so it sails through as a distinct payable. The system never sees them side by side until it is too late.

An agent matching payables to catch duplicates before the run
The duplicate check runs across every entry channel before the payment run, matching fuzzily rather than on an exact key.
02

Why exact matching does not catch them

The instinct is to block exact-duplicate invoice numbers, and that catches the easy cases and none of the hard ones. Real duplicates are usually fuzzy: the invoice number has an extra digit or a different prefix, the vendor name is spelled two ways, the amount differs by a rounding cent or a currency conversion, or the date is off by the entry lag.

A human reviewer would spot most of these by pattern, but no one has time to eyeball the whole payable population before every run. So the fuzzy duplicates are exactly the ones that get paid twice.


Exact-match blocks catch the easy duplicates and miss the fuzzy ones, which are the duplicates that actually get paid twice.
03

What the agent checks before the run

Pre-run duplicate check

An agent checks the full payable population before each payment run, matching on the combination that actually identifies a duplicate rather than on a single exact field.

Fuzzy vendor and invoice. Vendor names and invoice references matched across spelling variants, prefixes, and transpositions, not just exact strings.
Amount and date windows. Amounts matched within a tolerance and dates within an entry-lag window, so a rounding cent or a day's delay does not hide a duplicate.
Across every channel. The check spans email, portal, and manual entries together, so a duplicate that came in a different door still collides with the original.
“

A duplicate payment leaves in seconds and comes back over weeks, if at all. The whole game is catching it before the run, not detecting it in next month's audit.

Ashok Manthena
04

Catching it before the money leaves

Prevention, not recovery

The timing is the whole point. A duplicate caught before the run is a held payable and a two-minute review; the same duplicate caught after the run is a recovery project. Running the check as part of preparing the payment run, not as a monthly audit, is what turns detection into prevention.

Suspected duplicates are held with the two candidates shown side by side and the reason for the flag, so a person confirms or releases quickly. The genuine ones never pay; the false positives release without friction.

A suspected duplicate held for review with both candidates shown
Caught before the run, a duplicate is a two-minute review, not a weeks-long recovery project.
05

The takeaway for AP teams

Duplicate payments are not a discipline problem you can approve your way out of; they are a matching problem across separate entry channels, and matching is what people do not have time to do by hand before every run.

Let an agent run the fuzzy check across the whole payable population before each run, and the duplicates get held before the money leaves, with a person deciding the edge cases. The quiet leak closes without another approval layer slowing everyone down.

The process

Catch duplicates before the run, across every channel.

Duplicate payments are a fuzzy-matching problem across separate entry channels, which is why exact-match blocks and extra approvals miss them. An agent checks the full population before each run and holds the suspects.

The genuine duplicates never pay, the false positives release without friction, and a person decides the edge cases before the money leaves.

Get started

See duplicates caught before the run.

Bring a payable population. We will show the agent matching fuzzily across channels and holding the suspected duplicates for review before any payment leaves.

✓Watch fuzzy matching catch near-duplicate invoices
✓See suspects held with both candidates side by side
✓Leave with a duplicate-prevention plan and a realistic timeline
Request a demo
Confidential & private