Freight operations guide2026-09-27

Not every invoice that looks alike is the same invoice.

A vendor rarely emails one invoice at a time. A workshop or a carrier is more likely to send a batch scan — several real bills, often with sequential invoice numbers, sometimes dated the same day. A duplicate check tuned for single-invoice uploads will see that pattern and flag the whole batch as repeats. A duplicate check built for freight AP has to tell the difference before anyone approves the button that removes them.

1. 1

   Do not merge on similarity alone

   Two invoices that look alike — same carrier, close dates, similar totals — are not automatically the same bill. Documents stating different invoice numbers are not merged into one duplicate group, even when everything else about them matches.

2. 2

   Keep every copy in the group, not just one

   When documents do look like real duplicates, the group keeps the whole set rather than picking one survivor and hiding the rest. You choose which copies to keep, with all of them still visible to check against each other.

3. 3

   Make a "not a duplicate" decision stick

   Once you tell the system two invoices are not duplicates, that decision persists. It does not reappear as a flag the next time a similar pair is uploaded, which matters most for a carrier that reuses a numbering pattern every month.

4. 4

   Removing a copy should be recoverable

   A copy you remove from a duplicate group goes to Trash rather than being deleted outright. If the judgement call was wrong — it happens, especially on a rushed month-end batch — the invoice is still there to bring back.

5. 5

   Nothing should ever delete itself

   The duplicate signal is exactly that — a signal. It never auto-deletes anything. A person confirms every removal, which is the only safe default when the downside of a wrong guess is a real bill going unpaid.