Purchase order matching is the control that keeps accounts payable honest—and the step that quietly burns the most hours. Before an invoice gets paid, someone has to confirm that three documents agree: the purchase order, the goods receipt, and the supplier invoice. When all three line up, payment is routine. When they don't—a unit price is off by a few cents, a quantity is short, a PO number is missing—the invoice falls into an exception pile that someone chases down by email and phone.
That manual work is expensive. Ardent Partners puts the average cost to process a single invoice at $9.40, while best-in-class teams do it for $2.78. The difference is almost entirely automation. This guide walks through how to automate purchase order matching end to end in 2026—including the part most tools gloss over: what happens to the invoices that don't match.
What Is 3-Way Matching? Three-way matching is the verification AP runs before paying a bill. It compares the purchase order (what you agreed to buy), the goods receipt (what actually arrived), and the invoice (what the supplier billed). If the item, quantity, and price agree across all three within a set tolerance, the invoice is cleared for payment. If any field disagrees, it becomes an exception that needs investigation. Two-way matching skips the goods receipt and compares only the PO and invoice; three-way is the stricter control most finance teams use for physical goods.
Step 1: Capture Invoices From Every Channel Invoices arrive as emailed PDFs, EDI feeds, supplier-portal uploads, and the occasional scanned paper copy. The first job is getting all of them into one structured pipeline. With Symphona Flow , an inbound-email trigger or file-upload trigger picks up each invoice as it lands, and an AI text-extraction step reads the header and line items—supplier, PO number, quantities, unit prices, totals—into clean fields. No template-per-vendor setup, and no re-keying.
Step 2: Pull the PO and Goods Receipt From Your Existing Systems Matching only works if you can reach the source documents. Using the PO number on the invoice, Flow calls your ERP or procurement system over REST, SOAP, or a direct database connection to retrieve the matching purchase order, then pulls the related goods receipt to confirm the items were actually delivered. Symphona sits on top of your existing stack rather than replacing it, so you're reading live data from the systems of record you already trust—not maintaining a second copy of your procurement data.
Step 3: Run Purchase Order Matching Line by Line Against Tolerance Rules With all three documents in hand, the process compares them field by field: does the invoiced quantity match what was received, does the unit price match the PO, do the line items reconcile to the total? Tolerance rules decide what counts as a match—say, a price variance under 2% or $5, whichever is lower. Invoices that clear tolerance are approved and posted automatically. This is where the cost curve bends: best-in-class AP teams reach roughly 49% straight-through processing and clear invoices in about three days, versus more than seventeen for everyone else.
Step 4: Route Every Exception to the Right Person Here's what the typical AP tool waves away. Automation handles the clean invoices easily; the real work is the exceptions—and they're common. The Institute of Finance & Management notes that a single error in manual processing can cost up to $53 to fix, and exceptions are where those costs accumulate. Instead of dumping mismatches into an inbox, Flow creates a structured task in Symphona Serve , assigned to the right buyer or approver with role-based fields so a clerk sees what a clerk should see and a manager sees more. The original PO, receipt, and invoice travel with the ticket, so no one re-hunts for context.
Step 5: Auto-Resolve the Exceptions That Repeat Most exceptions aren't unique—they're the same handful of problems over and over: a missing goods receipt, a price the supplier changed without telling you, a quantity short-shipped. Symphona Resolve captures each failed match as a Fallout with full execution context, then lets you build resolution logic that handles the recurring cases without a human. If an invoice fails because the goods receipt hasn't posted yet, Resolve can wait and retry; if a price is off, it can email the supplier for a corrected invoice and resume the match when the reply lands. People handle the genuinely ambiguous cases; the routine ones resolve themselves. IOFM reports that automated three-way matching can cut exception rates by up to 60% versus manual operations.
Step 6: Monitor and Tune the Workflow Once matching runs on its own, the dashboard becomes your control panel. Track straight-through rate, average days to pay, and which suppliers or PO types generate the most exceptions. A vendor whose invoices fail half the time usually points to a catalog or pricing problem upstream—fixing it removes the exceptions at the source rather than processing them faster. And because every action is logged from invoice capture through payment, you get an end-to-end audit trail that satisfies internal finance controls and external auditors alike.
The Bottom Line Automating purchase order matching isn't just running OCR on invoices. The full workflow is capture, retrieval, line-level matching against tolerance, intelligent exception routing, and automated resolution of the cases that repeat. Tools that stop at "we read the invoice" leave your team doing the expensive part by hand. A platform that handles both the clean matches and the messy exceptions in one place is what moves an AP function from 17 days and $9.40 an invoice toward best-in-class.
Manufacturers and other high-volume buyers feel this most—thousands of POs, receipts, and invoices a month, each one a chance for a mismatch. If your team is ready to stop chasing exceptions by email, explore how SimplyAsk.ai approaches automation for manufacturing operations , or book a consultation to map your own three-way matching workflow end to end.