Credit and rebill pairing: the correction that reads as two errors

When a supplier fixes a mistake, they rarely edit the original invoice. They issue a credit memo that reverses the line, then a new invoice that bills it correctly. Two documents, three lines, one economic event. An audit that treats each line independently reports a large overcharge on the original, a large negative variance on the credit, and a small or zero variance on the rebill. All three numbers are wrong as a description of what happened.

Why it matters more than it sounds

Unpaired credits inflate both sides of your variance report. Your recovery pipeline fills with claims the supplier already settled, which is the fastest way to lose credibility in a dispute conversation. Meanwhile genuine overcharges sit further down the list because the phantom ones outrank them.

What a pair actually looks like

• Same supplier, same SKU, opposite sign on quantity or extended amount.
• The credit references the original document number — sometimes in a dedicated field, often only inside the description text.
• The credit date is later than the original, usually inside 60 days but occasionally much longer.
• The rebill may split one original line into several, or consolidate several into one.
• Quantities may not match exactly when the correction is partial.

A pairing sequence that survives real data

• First pass: match on an explicit original-document reference where the supplier provides one. This is the only match you can trust without further checks.
• Second pass: match on supplier plus SKU plus equal-and-opposite quantity within a date window, nearest first.
• Third pass: allow partial quantity matches, consuming the original line's quantity as credits are applied so a line cannot be credited twice.
• Extract document references from free-text description fields before giving up — a surprising share of credits identify their parent only in prose.
• Leave anything still unpaired as an explicit 'unpaired credit' reason rather than silently netting it against the period.

Then classify the event, not the lines

Once paired, the audit question becomes simple: after the credit and rebill, did the net price for that item on that date match the contract? If yes, there is nothing to claim and nothing to report beyond a note that a correction occurred. If no, the residual difference is the real variance, and it is the only number worth putting in front of a supplier.

Keep the unpaired credits visible

An aging list of unpaired credits is a useful artefact in its own right. A credit issued and never followed by a rebill can mean money you were owed and never received, or a return that was never re-invoiced. Both deserve a look, and neither shows up if you quietly net credits into the monthly total.