Three-Way Matching: How It Works in AP

What Three-Way Matching Is

Three-way matching is the accounts payable control that confirms an invoice is legitimate by comparing it against two other documents — the purchase order and the goods receipt — before any payment is made. The principle is simple and powerful: a company should only pay for what it actually ordered and actually received, at the price it agreed. When the three documents tell the same story, the invoice is safe to pay.

This control sits inside the broader procure-to-pay process and is one of the seven core procurement process steps. This guide covers the foundational mechanics; for how modern software automates the work, see our companion analysis of AI-powered three-way matching, which benchmarks what automation actually achieves.

The Three Documents

Each document answers a different question, and the match only holds when all three answers agree.

• Purchase order (what was ordered): the binding commitment issued to the supplier, carrying agreed quantities and prices.
• Goods receipt note (what was received): receiving's record of what physically arrived, by quantity and often condition. See our explainer on the goods receipt note.
• Supplier invoice (what is billed): the supplier's request for payment, which must reconcile to both of the above.

How Three-Way Matching Works, Step by Step

1. The invoice arrives and its PO number is identified.
2. The matching engine retrieves the referenced purchase order.
3. It retrieves the corresponding goods receipt.
4. It compares quantity, unit price, total, and line descriptions across all three.
5. If everything agrees within tolerance, the invoice is cleared for payment.
6. If anything falls outside tolerance, the invoice is flagged as an exception for review.

In a manual environment, an AP clerk performs every step by hand, which typically takes several minutes per invoice and carries a meaningful error rate. The whole point of automation is to do this comparison instantly and consistently across thousands of invoices.

Two-Way vs Three-Way vs Four-Way Matching

Matching comes in degrees. Each added document strengthens control at the cost of more handling. The right level depends on the risk and value of the spend.

For a focused comparison of the two most common methods, see two-way vs three-way matching and the related breakdown of 3-way match vs 2-way match.

Setting Matching Tolerances

Tolerances define how close the documents must be before an invoice clears automatically. A small price or quantity variance is often legitimate — rounding, a minor freight charge, an agreed quantity flex — and forcing every one of these to a human wastes time. Tolerances let routine variances pass while genuine discrepancies get caught.

The trade-off is real. Tight tolerances catch more issues but generate more exceptions and slower processing. Loose tolerances clear more invoices but let small leakages through. Most organizations tune tolerances by spend category and supplier, applying stricter rules where the financial exposure justifies the friction.

Common Matching Exceptions

Exceptions are not failures; they are the system doing its job. The usual culprits are partial shipments, where the invoice covers more than the receipt yet recorded; price variances between PO and invoice; quantity variances from short or consolidated shipments; missing receipts where the invoice simply arrived first; and brand-new suppliers with no PO on file. Each exception type needs a defined resolution path, and the share of invoices that become exceptions is a direct read on data quality upstream. Our invoice matching guide goes deeper on handling each category.

Benefits and Why the Control Exists

Three-way matching exists because the alternative is expensive. Without it, companies pay duplicate invoices, settle bills for goods that never arrived, and absorb price creep no one notices. With it, every payment is tied to a verified order and delivery, fraud has fewer places to hide, and the audit trail is automatic. It is the single most important financial control in accounts payable, which is why even highly automated teams keep it at the center of the process.

A Worked Matching Example

Consider a purchase order for 100 units of a component at $10 each, a total of $1,000. The supplier ships in two batches: 60 units arrive first, recorded on a goods receipt, then 40 units a week later on a second receipt. The supplier invoices $1,000 for all 100 units once both shipments are complete.

The matching engine checks the invoice against the PO ($1,000 ordered) and the combined goods receipts (100 units received). Quantities, prices, and totals agree, so the invoice clears. Now change one detail: the invoice bills 100 units but only 90 were received. The match fails on quantity, and the invoice is flagged as an exception. AP investigates, finds that 10 units were back-ordered, and either holds the invoice or pays for the 90 received and matches the balance later. This is the control working as designed — the company does not pay for goods it has not received.

The example also shows why partial shipments are the most common exception: the timing of receipts and invoices rarely lines up perfectly. A good matching process expects this and handles it gracefully rather than treating every partial as a problem. Our invoice matching guide details how each exception type is resolved, and the purchase order process explainer covers how clean POs prevent exceptions in the first place.

When Four-Way Matching Makes Sense

Four-way matching adds a fourth document to the comparison — an inspection or quality certificate — confirming not just that goods arrived but that they passed quality control. It is overkill for routine purchases but essential where receiving defective goods carries real consequence: pharmaceuticals, aerospace components, food ingredients, and other regulated or safety-critical categories.

The trade-off is the same as always, intensified. Each added document strengthens the control and slows the process, so four-way matching should be reserved for the spend that genuinely warrants it. Most organizations run three-way matching as the default and reserve four-way for a defined list of critical categories. Deciding which categories qualify is a risk judgment best made with the same logic as the Kraljic Matrix uses to separate critical spend from routine.

Best Practices for Clean Matching

High match rates are earned upstream, not at the moment of matching. The single most important practice is PO discipline: complete, accurate purchase orders with correct line items, quantities, and prices give the engine something reliable to match against. The second is accurate, timely receipting — the receipt is one of the three documents, and sloppy receiving guarantees exceptions.

Beyond data, tune tolerances by category and supplier so routine variances clear automatically while material ones are caught. Enforce no-PO-no-pay so every invoice has a PO to match against. And track your exception rate over time — a rising rate is an early warning that data quality is slipping somewhere upstream. Teams that treat the exception rate as a managed metric, rather than a fact of life, steadily push it down. As our analysis of automated matching repeatedly finds, the gap between a 65% and an 85% match rate is almost always data quality, not the sophistication of the engine. The accounts payable process guide places these practices in the wider AP context.

Common Matching Mistakes

The recurring mistakes are predictable. Setting tolerances too loose clears invoices that should be questioned, letting small overcharges accumulate. Setting them too tight floods AP with false-positive exceptions, burning staff time on invoices that were fine. Allowing non-PO invoices breaks matching entirely, since there is nothing to match against. And ignoring receipt accuracy shifts the failure point to receiving, where it is harder to see.

The deeper mistake is treating matching as an AP-only concern. Because the inputs — POs and receipts — are created by procurement and receiving, the match rate is really a measure of the whole organization's data discipline. Fixing it means working across teams, not just tightening the AP rulebook.

Automation and AI in Matching

Automated matching turns a slow, manual comparison into an instant one. Rules-based engines apply your tolerances; learning-based engines go further, recognizing from history which variances are routine for a given supplier and which warrant a second look. The result is that the bulk of invoices clear without a human touching them, and staff time shifts from data entry to resolving the exceptions that genuinely need judgment.

Our benchmark work, summarized in the AI three-way matching analysis, found that real-world automated match rates usually sit in the 75–90% range, driven more by PO and receipt data quality than by the sophistication of the engine. To compare platforms that do this well, review the invoice and AP automation tools and reviews like Stampli, then browse the wider blog for related back-office automation topics.

Three-Way Matching in Your ERP

In most organizations, three-way matching runs inside or on top of the ERP, because that is where the purchase order, goods receipt, and vendor data already live. Native ERP matching applies your configured tolerances automatically and blocks invoices that fail, holding them for review. The strength of native matching is that it works directly against the system of record; its limitation is that traditional ERP matching is rigid — it applies fixed rules and cannot learn which variances are routine for a given supplier.

This is the gap that dedicated AP automation and AI matching tools fill. They sit alongside the ERP, ingest invoices in any format, and apply more flexible, learning-based matching before writing clean results back. The decision between native ERP matching and an add-on platform comes down to volume and complexity: organizations with high invoice volumes, many suppliers, and messy formats usually justify a dedicated tool, while those with modest, clean volumes may do well with native capability. Either way, the integration between the matching layer and the ERP is decisive, a point our procure-to-pay process guide stresses repeatedly.

The Cost of Manual vs Automated Matching

The business case for automating three-way matching rests on a simple contrast in handling cost. Manual matching requires an AP clerk to retrieve and compare three documents for every invoice — typically several minutes each, with a meaningful error rate. Across thousands of invoices a month, that adds up to a substantial share of the AP team's time spent on repetitive comparison rather than judgment.

Automated matching inverts the ratio. The engine clears the majority of invoices without human involvement, and staff handle only the exceptions — the partial shipments, price variances, and missing receipts that genuinely need a decision. Our analysis suggests this typically shifts the AP team from reviewing every invoice to reviewing a minority of them, freeing well over half of the manual effort for higher-value work. The realized saving depends heavily on data quality: clean PO and receipt data drives high match rates and low exception volumes, while dirty data erodes the benefit. To estimate the value for your own volumes, our ROI calculator provides a structured starting point, and the source-to-pay market analysis shows where the automation tools sit.

The Bottom Line on Three-Way Matching

Three-way matching endures because it answers the one question every payment must satisfy: did we order this, did we receive it, and is the price right? No amount of automation changes that logic; it only changes how much of the checking a human has to do. The control is as relevant in a fully automated accounts payable function as it was in a paper-based one.

For practitioners, the priorities are clear. Get the upstream data right — clean purchase orders and accurate receipts — because that is what determines match rates far more than any engine. Tune tolerances to your spend, so routine variances clear automatically and material ones are caught. Reserve four-way matching for the critical categories that warrant it. And when you automate, weigh integration depth and exception handling at least as heavily as headline accuracy claims, which our benchmark analysis shows are routinely overstated. Done well, matching shifts from a manual chore to a near-invisible control that protects every dollar while freeing the team for work that needs judgment. To compare the platforms that automate it, start with the invoice and AP automation tools and the broader procure-to-pay process guide.

Frequently Asked Questions

What is three-way matching?

Three-way matching is an accounts payable control that compares three documents before an invoice is paid: the purchase order (what was ordered), the goods receipt (what was received), and the supplier invoice (what is being billed). When the quantities, prices, and descriptions agree within tolerance, the invoice is approved for payment.

What are the three documents in three-way matching?

The three documents are the purchase order, the goods receipt note, and the supplier invoice. The PO establishes what was agreed, the goods receipt confirms what physically arrived, and the invoice states what the supplier is charging. Matching all three protects the company from paying for goods it did not order or receive.

What is the difference between two-way and three-way matching?

Two-way matching compares only the invoice and the purchase order, so it confirms the bill matches the order but not that goods were received. Three-way matching adds the goods receipt, confirming delivery as well. Four-way matching adds an inspection or quality certificate. More documents mean stronger control but more handling.

What is a matching tolerance?

A tolerance is the allowable difference between documents before an invoice is flagged as an exception. For example, a 2% price tolerance lets minor variances pass automatically. Tolerances balance control against efficiency: tight tolerances catch more issues but create more exceptions, while loose tolerances clear more invoices but let small discrepancies through.

How does AI improve three-way matching?

AI matching learns from historical successful matches to clear standard invoices automatically and to recognize which variances are routine versus genuinely problematic. It handles partial shipments, expected price differences, and timing gaps better than rigid rules, routing only true exceptions to staff. Our benchmark analysis found real-world automated match rates typically land in the 75–90% range depending on data quality.