What a Posted Goods Receipt Actually Proves

Of the three documents in a three-way match, the posted goods receipt is the one carrying the most evidential weight — and the one people think about least.

Why the receipt is strong evidence

Three properties, and each matters:

It is made by someone else. The person who counts the delivery is not the person who raised the order and not the person who processes the invoice. Separation of duties is not a formality here; it is why the receipt is worth anything.

It is made before the invoice arrives. The count is not influenced by what the vendor later says arrived. Nobody is reconciling towards a number they have already seen.

It is made against physical reality. Somebody looked at the pallet. Whatever its faults, the receipt is a record of an observation, not of an intention or an assertion.

Compare that to the invoice, which is the counterparty’s claim about a transaction they benefit from, and to the order, which is a statement about the future made before anything happened.

Uninvoiced received quantity

This is the figure that actually matters, and it is not simply “how much arrived”.

For any purchase order line, take every posted receipt line against it and sum:

received quantity − quantity already invoiced

That is the uninvoiced received quantity: how much has physically arrived and has not yet been billed for.

It is the correct figure because it handles the two things that make real deliveries messy:

• Partial deliveries. An order for 100 arriving as 60 then 40 produces two receipt lines. Both count.
• Partial invoicing. A vendor who bills for the first 60 and then, later, for all 100, is trying to bill you twice for the same 60. Tracking what has already been invoiced is what catches that.

An invoice claiming more than the uninvoiced received quantity is claiming for goods that either did not arrive, or have already been paid for. Both are worth stopping.

Why quantity tolerance should be zero

Price tolerance makes sense. Prices legitimately move a little — rounding, small currency effects, a minor agreed adjustment.

Quantity does not work that way. Either 5 units arrived or 8 did. There is no “approximately six and a half units” that a sensible tolerance band accommodates. A quantity discrepancy is always a real event: a short delivery, a miscount, a double bill, or a vendor error.

That is why a well-designed system treats quantity as exact and configures tolerance only on price and tax. A quantity tolerance does not absorb noise, because there is no noise to absorb. It only absorbs findings.

The asymmetry

A good check is one-sided in your favour:

• Claiming more than the uninvoiced received quantity stops the document.
• Claiming less is a warning, not a failure. Being billed for less than arrived is not a loss, and it is often perfectly legitimate — a vendor billing in instalments against a single delivery.
• Charging less than the agreed price is always accepted, at any magnitude.
• Charging more is accepted only within tolerance.

You are checking for overcharge, not for perfect symmetry. Building the checks to reflect that keeps the exception queue focused on things worth a person’s attention.

What a receipt cannot tell you

Honesty about the limits matters as much as the strengths.

It cannot tell you the goods were correct. A receipt records that a quantity of an item arrived. Whether they were the right specification, undamaged, or fit for purpose is a quality question, and quality processes sit elsewhere.

It cannot tell you the price was right. That is the order’s job.

It cannot help with anything that has no receipt. Services, professional fees, rent, utilities, one-off repairs — none of these produce a goods receipt. There is nothing physical to count.

That last one is the real boundary. Any system whose safety argument rests on the posted receipt has no such argument for spend that never produces one. The honest response is to say so, and to require a different control — a human coding the expense and an approver authorising it — rather than to pretend the same guarantee applies. See PO-backed and non-PO spend.

Why receipts get posted late, and what it costs

The single most common cause of an invoice sitting in an exception queue is not fraud, and not a vendor error. It is that the receipt has not been posted yet.

The goods arrived. The invoice arrived. The paperwork on the dock has not caught up. The invoice cannot be reconciled against a receipt that does not exist, so it stops.

The fix is operational, not technical: post receipts promptly. The payoff is disproportionate — an organisation that posts receipts the day goods arrive gets a mostly-empty exception queue, and one that posts them weekly gets a queue full of documents that are not actually wrong.

If you are evaluating invoice automation and your receipts run days behind, expect that to be the first thing you notice, and expect it to be the cheapest thing to fix.