Three-Way Matching
Three-way matching is an accounts payable control that compares three documents before approving an invoice for payment: the purchase order (what was ordered), the goods receipt (what was delivered), and the supplier invoice (what is being billed). If the quantities, prices, and terms across all three documents agree within defined tolerances, the invoice is approved automatically. Discrepancies — wrong quantities, unexpected price changes, missing line items — are flagged as exceptions for human review. Three-way matching is the most widely used AP control because it catches the most common payment errors: paying for goods not received, paying the wrong price, and paying duplicate invoices.
- Three-way matching compares three documents: the purchase order, the goods receipt, and the supplier invoice
- The goal is to verify that what was ordered, what was received, and what was billed all agree before payment
- Manual three-way matching takes 5–15 minutes per invoice and is the #1 bottleneck in accounts payable
- Automated three-way matching completes in seconds with 98% accuracy, flagging only genuine exceptions for review
- Four-way matching adds the contract or price agreement as a fourth document, catching unit price overcharges
- Organizations using automated matching report 90%+ reduction in duplicate payments and pricing errors
How Three-Way Matching Works
The three-way matching process compares specific data points across three documents:
Purchase order (PO)
The original order document, containing item descriptions, quantities, unit prices, delivery terms, and the PO number. This is the buyer's commitment to purchase.
Goods receipt (GR)
The record created when goods arrive at the buyer's location. The receiving team logs what was actually delivered — quantities, item descriptions, condition. Partial deliveries create partial goods receipts against the same PO.
Supplier invoice
The bill sent by the supplier requesting payment — see purchase order vs invoice for the distinction. It should reference the PO number and list the same items, quantities, and prices as the PO.
The match logic: - PO quantity ≥ GR quantity ≥ Invoice quantity (you shouldn't be billed for more than you received or ordered) - PO unit price = Invoice unit price (within tolerance — typically 1–2%) - Invoice total = sum of line items × unit prices + tax - PO number on invoice matches an open PO in the system
When all conditions pass, the invoice is approved for payment without human intervention. When any condition fails, the invoice is routed to an exception queue with context about what didn't match and why.
Three-Way vs. Four-Way Matching
Three-way matching catches quantity and delivery discrepancies. Four-way matching adds a fourth document — the contract or price agreement — to catch pricing discrepancies that three-way matching misses.
When three-way matching is sufficient: Standard purchases with stable pricing, catalog orders, repeat purchases from established suppliers where prices rarely change.
When four-way matching is necessary: Complex procurement with negotiated pricing, volume discounts, or tiered pricing. The contract specifies the agreed unit price; the PO may not reflect promotional pricing or retroactive adjustments. Four-way matching catches cases where the invoice price doesn't match the contractual price — even if it matches the PO.
Example: Your contract says $10/unit. The PO was accidentally created at $12/unit. The supplier invoices $12/unit. Three-way matching passes (PO = Invoice). Four-way matching catches it (Contract ≠ Invoice).
Organizations handling complex procure-to-pay workflows with negotiated supplier agreements should implement four-way matching from day one. The incremental effort is minimal; the protection is significant.
The Problem with Manual Matching
Manual three-way matching is the single biggest bottleneck in accounts payable. Here's why:
Time per invoice
An AP clerk manually matching one invoice against a PO and goods receipt spends 5–15 minutes: pulling up the PO, comparing line items, checking the GR, verifying quantities and prices, documenting the match. For an AP team processing 5,000 invoices monthly, that's 400–1,250 hours of matching work alone.
Error rates
Human matching produces 2–5% error rates. At scale, that means dozens of overpayments, duplicate payments, or missed discrepancies every month. Each error costs $50–$200 to investigate and correct — a key driver of AP automation ROI.
Exception handling
When a match fails, the clerk has to investigate: Was there a partial delivery? Did the supplier change the price? Is this a duplicate? Manual investigation takes 20–45 minutes per exception and often requires contacting the supplier.
Automating Three-Way Matching
Automated three-way matching eliminates the manual effort by using software to extract data from all three documents and compare them algorithmically:
Document capture
The system ingests invoices from any channel (email, PDF upload, supplier portal, EDI). AI extracts line-item data: item descriptions, quantities, unit prices, PO references, tax amounts.
Confidence scoring
Each match result gets a confidence score. Invoices scoring above threshold (typically 90%) auto-post to the ERP. Those in the review band (50–90%) route to AP staff with full match context. Below 50% escalate immediately.
The result: 85–98% of invoices process without human touch. Your AP team focuses on the 2–15% that genuinely need judgment — not on data entry and document lookup. PO automation works in tandem with matching automation to keep the upstream PO data clean and current.
Common Three-Way Matching Exceptions
Understanding the most frequent exception types helps AP teams resolve them faster and, more importantly, fix the root causes that generate them:
Quantity mismatch (GR ≠ Invoice)
The supplier invoices for 500 units but the goods receipt shows only 450 received. This is the most common exception. Causes: partial shipment not yet recorded, receiving team hasn't logged the delivery, or the supplier invoiced before all shipments arrived. Resolution: check for pending deliveries against the PO, then accept or reject the variance.
Price variance (PO ≠ Invoice)
The PO says $10.00 per unit but the invoice says $10.50. Causes: the supplier applied a price increase not yet reflected in the PO, currency conversion differences on international orders, or the PO was created from an outdated price list. Resolution: verify against the contract or latest supplier price sheet, amend the PO if the new price is legitimate, or reject the invoice back to the supplier.
Missing goods receipt
The invoice arrives but no goods receipt exists in the system. The goods may have been delivered but not logged, or the invoice arrived before the shipment. This exception blocks the match entirely. Resolution: contact the receiving team to verify delivery status. For service-based POs, a service acceptance confirmation replaces the goods receipt.
No PO reference
The invoice doesn't include a purchase order number, making automatic matching impossible. Causes: the purchase was made without a PO (maverick spend), or the supplier omitted the PO reference. Resolution: route to the requester for PO identification or escalate as unauthorized spend.
Duplicate invoice
The same invoice number from the same supplier appears twice. Causes: the supplier sent the invoice via both email and postal mail, or a revised invoice was sent without voiding the original. Automated matching systems detect duplicates by checking invoice number + supplier ID + amount combinations.
Unit of measure mismatch
The PO specifies "cases" but the invoice bills by "units." The total amount may be correct, but the line-item comparison fails because the UOM doesn't match. Smart matching engines handle UOM conversion tables automatically; legacy systems flag these as exceptions.
Three-Way Matching Best Practices
Organizations that achieve the highest match rates and lowest exception volumes follow these practices:
Set tolerances thoughtfully
Zero-tolerance matching generates excessive exceptions from rounding differences and minor shipping variances. Start with 1–2% on price and 1–5 units on quantity, then adjust based on your actual exception data. Different commodity categories and suppliers may warrant different tolerances — a 2% variance on a $5 item is very different from 2% on a $50,000 line item.
Fix root causes, not symptoms
When the same exception type recurs for the same supplier, don't just resolve each instance individually. Investigate: is the supplier consistently invoicing at a different price? Is the receiving team not recording partial deliveries? Fixing the upstream problem eliminates entire categories of exceptions permanently.
Automate goods receipt recording
A surprising number of matching failures trace back to missing or delayed goods receipts. Integrate receiving workflows with your ERP so that barcode scans or dock-door confirmations automatically create goods receipt records. The faster the GR appears in the system, the fewer "missing GR" exceptions your AP team has to investigate.
Implement pre-match validation
Before invoices enter the matching queue, validate that mandatory fields are present and correctly formatted: PO number exists in the system, supplier ID matches the PO vendor, line items are parseable. Rejecting malformed invoices at the point of entry prevents them from clogging the exception queue.
Track match rate by supplier
Some suppliers consistently generate exceptions due to billing practices, format inconsistencies, or price-change communication gaps. Identifying your top-10 exception-generating suppliers and working with them to improve invoice quality can reduce your total exception volume by 30–50%.
Review tolerance settings quarterly
As your automation learns and your supplier relationships mature, tolerance settings should evolve. Tighten tolerances where exception rates are low to catch more issues; loosen them where frequent false positives waste reviewer time.
Your operations, on autopilot.
GeneralMind handles procure-to-pay and order-to-cash end-to-end — 98% decision accuracy, full auditability, zero manual steps. See it live in 30 minutes.
Book a demoHow GeneralMind Automates Matching
GeneralMind runs three-way and four-way matching automatically on every invoice it processes. Our solution extracts line-item data from supplier invoices with 98% accuracy using automated invoice processing, pulls the corresponding PO and goods receipt data from your ERP in real time, and compares everything at the line-item level.
Configurable tolerance thresholds let you control sensitivity per vendor, commodity, or business unit. The matching engine handles complex real-world scenarios out of the box: partial deliveries across multiple goods receipts, unit-of-measure conversions, multi-currency invoices, and retroactive price adjustments. Invoices that match cleanly auto-post to your ERP. Exceptions route to reviewers with full context — what didn't match, by how much, and what the likely cause is.
The system learns from every correction, continuously improving match accuracy and reducing exception volumes over time. Clients typically see first-pass match rates climb from 80% in month one to 95%+ by month six — without manual rule tuning. Native integrations with SAP, NetSuite, Dynamics 365, Sage, and Infor mean PO and GR data is always current — no stale extracts, no reconciliation gaps.
Frequently Asked Questions
Three-way matching is an accounts payable control that compares the purchase order, goods receipt, and supplier invoice before approving payment. It verifies that what was ordered, delivered, and billed all agree.
Three-way matching compares PO, goods receipt, and invoice. Four-way matching adds the contract or price agreement as a fourth document, catching unit price discrepancies that three-way matching misses.
Manual three-way matching takes 5–15 minutes per invoice, including document lookup, line-item comparison, and exception investigation. For high-volume AP teams, this adds up to hundreds of hours monthly.
PO matching automation uses software to automatically extract invoice data, retrieve the corresponding PO and goods receipt from your ERP, and compare them algorithmically. Clean matches auto-post; exceptions route to reviewers with context.
Typical tolerances are 1–2% on unit price and 1–5 units on quantity. Start conservative (tight tolerances), then loosen based on your actual exception patterns. The goal is to auto-approve clean invoices while catching genuine discrepancies.
The most common causes are: quantity mismatches between the goods receipt and invoice (partial deliveries), price variances between the PO and invoice (unapproved price changes), missing goods receipts (delivery not yet logged), invoices without a PO reference (maverick spend), and duplicate invoices submitted via multiple channels.
Two-way matching compares only the purchase order and the invoice — skipping the goods receipt. It verifies that the invoice matches what was ordered, but not what was delivered. Two-way matching is simpler and faster but offers weaker fraud protection since it doesn't confirm that goods were actually received before payment.
Three-way matching prevents several fraud types: fictitious invoices (no corresponding PO or goods receipt), inflated invoices (quantities or prices exceed the PO), and collusion between a buyer and supplier (the goods receipt provides independent verification from the receiving team). Without the goods receipt check, fraudulent invoices backed by a legitimate PO could pass undetected.
Manual three-way matching typically produces exception rates of 20–30% — meaning one in four or five invoices requires manual investigation. Automated matching reduces this to 2–15% by handling tolerance-based variances, UOM conversions, and partial deliveries automatically. Best-in-class organizations achieve exception rates below 5%.
