Two invoice-matching methods sit at the heart of accounts payable control. The difference comes down to one document — the goods receipt — and that single document decides how much fraud and error you catch, how fast invoices clear, and which spend each method belongs on.
Last updated: · Reviewed by Fredrik Filipsson
The two methods share a goal — pay only what you truly owe — but differ in how far they verify a transaction before releasing payment.
One document separates them. Two-way matching reconciles the supplier invoice against the purchase order. Three-way matching inserts the goods receipt note as a third checkpoint, so the invoice must agree with proof of delivery as well as the order.
Control versus speed is the trade-off. The receipt step in three-way matching is exactly what catches phantom deliveries, over-billing, and short shipments — but it also slows the cycle because a receipt has to be logged before payment can release.
It is not an either-or decision. In our analysis, mature AP functions run a tiered policy: three-way matching for inventory and capital goods, two-way matching for services and recurring spend below a set threshold. Read our wider view in the source-to-pay AI market analysis.
The lighter-weight control: two documents, one comparison.
Two-way matching is the practice of comparing a supplier invoice directly against the purchase order it relates to before approving payment. The accounts payable team — or, increasingly, the AP system — checks that the unit price and quantity on the invoice line up with what was authorised on the PO, within whatever tolerance the organisation allows for rounding, freight, or minor price drift.
Because only two documents are involved, the method is fast. There is no waiting for a warehouse or requester to confirm that something arrived. As soon as the invoice lands and the PO is on file, the comparison can run. That speed is precisely why two-way matching is common for spend categories where the idea of a goods receipt does not apply: a software subscription renewal, a monthly utility bill, an office lease payment, or a professional services retainer. In those cases there is no physical delivery to confirm, so a receipt document would add friction without adding assurance.
The control it provides is real but bounded. Two-way matching confirms that you are being billed for what you agreed to buy at the price you agreed to pay. What it cannot confirm is whether you actually received it. An invoice can match a purchase order perfectly while the goods sit undelivered — or were never shipped at all. That gap is the central limitation, and it is the reason higher-risk spend usually demands a third document. To understand where two-way matching sits in the wider cycle, it helps to see how the purchase order process feeds the documents this check relies on.
The control standard for goods: three documents must agree.
Three-way matching closes the gap that two-way matching leaves open. It compares three documents before any invoice is cleared for payment: the purchase order, which records what was authorised and at what price; the goods receipt note, which records what was physically delivered and accepted; and the supplier invoice, which records what is being billed. Only when all three agree on quantity and price — within tolerance — does the invoice pass.
The addition of the goods receipt is what makes the method robust. It forces a positive confirmation that the organisation received what it is about to pay for. A supplier cannot bill for ten units when only eight arrived; a duplicate or inflated invoice will fail against the receipt; and an entirely fictitious delivery has no receipt to match against at all. This is why three-way matching is the long-standing standard for inventory, raw materials, capital equipment, and any high-value physical purchase where the cost of paying for something you never received is significant.
The price of that assurance is process. Someone — or some system — has to log the receipt accurately and promptly, and the invoice cannot settle until they do. When receipts lag behind deliveries, invoices pile up in an exception queue, early-payment discounts get missed, and suppliers chase payment. The control is only as strong as the receipting discipline behind it, which is one reason automation has become central to making three-way matching work at scale. Our deeper walkthrough of the mechanics lives in the guide to AI three-way matching of invoices and purchase orders.
Side by side, the contrast is sharpest across documents, control, exceptions, speed, and best-fit spend.
| Dimension | Two-Way Matching | Three-Way Matching |
|---|---|---|
| Documents compared | Purchase order + supplier invoice | Purchase order + goods receipt + supplier invoice |
| What it controls against | Price and quantity billed vs what was ordered | Price and quantity billed vs both what was ordered and what was received |
| Confirms goods received? | ✗ No — receipt is not checked | ✓ Yes — receipt is the third checkpoint |
| Typical exception types | Price variance, quantity variance, missing PO | Price variance, quantity variance, short or partial delivery, missing receipt, over-billing |
| Speed to pay | Faster — no dependency on a receipt | Slower — invoice waits on receipt being logged |
| Audit & fraud strength | Moderate — catches billing errors, not phantom deliveries | Strong — catches phantom and inflated deliveries |
| Best-fit spend type | Services, subscriptions, utilities, rent, low-value indirect | Inventory, raw materials, capital goods, high-value physical spend |
Key takeaway: the goods receipt row is the whole story — every other difference flows from whether you verify delivery before paying.
Comparing AP automation tools that run matching for you? See our invoice and AP automation category.
Browse Invoice & AP AIA stricter cousin used where inspection and acceptance matter.
Some organisations go a step further with four-way matching, which adds a fourth document to the chain: an inspection or quality-acceptance record. Here the invoice must agree with the purchase order, the goods receipt, and confirmation that the delivered items passed inspection and were accepted as fit for use. It is most common in manufacturing, pharmaceuticals, aerospace, and other settings where receiving an item is not the same as accepting it — a shipment can arrive in full yet fail a quality check.
Four-way matching is the strongest control of the family and, predictably, the slowest and most resource-intensive. For most general procurement it is overkill; three-way matching captures the great majority of risk. It earns its place only where quality acceptance is a genuine gating step before payment should be considered legitimate.
The practical answer is a policy, not a single choice. Route spend to the right method by category and value.
There is no physical delivery to confirm — utilities, software subscriptions, rent, insurance, professional retainers. Also for low-value indirect purchases below a defined threshold, where the cost of a receipt step outweighs the risk it mitigates.
You are buying physical goods, especially inventory, raw materials, and capital equipment, or any high-value purchase above your threshold. The receipt confirmation is the control that protects against paying for goods that never arrived or arrived short.
You run mixed spend — which is almost everyone. Define a dollar threshold and a category list so the AP system routes each invoice to the right method automatically, keeping low-risk invoices fast and high-risk invoices tightly controlled.
Where matching sits in the broader buying cycle also shapes the policy. If you are mapping these controls across procurement, our comparison of source-to-pay vs procure-to-pay shows how invoice matching fits the wider process, and the invoice and AP AI category covers the tools that enforce it.
Modern AP platforms turn matching from a manual reconciliation chore into an exception-only workflow.
Manual matching is slow and error-prone: a clerk pulls up the invoice, finds the linked purchase order, hunts for the goods receipt, and compares line by line. AI-driven AP automation collapses that work. The system captures invoice line data on arrival, retrieves the corresponding PO and goods receipt automatically, and runs the comparison in seconds. Invoices that match cleanly within tolerance clear without a human ever touching them; only genuine exceptions surface for review.
The harder value is in how AI handles messy reality. Real invoices rarely line up cleanly with POs and receipts — units of measure differ, deliveries arrive in partial shipments, and suppliers consolidate several orders onto one invoice. AI interprets these the way an experienced clerk would, reconciling a "case of 12" against "12 each" or splitting a consolidated invoice back across its source orders. In our analysis, a well-configured system auto-clears a large majority of invoices and routes only the true exceptions to staff, which is where the labour savings and faster cycle times come from. Tools such as Tipalti and Stampli build their AP workflows around exactly this kind of automated matching.
Automation also changes the speed-versus-control calculus described earlier. The historic reason to favour two-way matching was that three-way matching was slow; when receipts are captured digitally and matched instantly, much of that friction disappears. That makes three-way matching practical for a wider range of spend than it used to be — though the underlying principle of routing by category and value still holds.
This is not a contest with a single winner. Two-way and three-way matching solve the same problem — paying only what you genuinely owe — at different levels of assurance, and each is the right tool for different spend.
Three-way matching is the stronger control and should be the default for physical goods, inventory, and high-value purchases. The goods receipt step is the only thing standing between you and paying for a delivery that never happened, and for that spend the friction is worth it. Two-way matching earns its keep on services, subscriptions, utilities, and low-value indirect spend, where a receipt is either impossible or pointless and speed matters more than an extra checkpoint.
The strongest AP functions in our analysis do not pick one. They write a tiered policy — three-way above a threshold and for goods, two-way below it and for services — and let an automation platform enforce the routing. Done well, that combination delivers tight control where risk is concentrated and fast throughput everywhere else.
Matching methods are tools, not absolutes. Map your spend categories first, set thresholds that reflect your real risk, and choose the platform that can enforce the policy automatically.
Common questions from AP managers and controllers setting their matching policy.
Continue building your accounts payable control knowledge.