How Purchase Order Numbers Work, From Requisition to Payment
A purchase order number gets created the moment a buyer commits to an order, well before any goods arrive or any invoice is issued. From that point on, it is the thread that connects the requisition, the approval, the goods receipt, and eventually the supplier's invoice. Accounts payable teams live and die by whether that thread stays intact, because a PO number that does not match anywhere along the chain is exactly what a three-way match check is designed to catch.
A purchase order number is a unique identifier assigned when an order is placed, typically formatted as a short prefix plus year plus running number (PO-2026-04831). Most accounting systems expect it to be sequential so gaps can be audited, though a purchase order generator is useful for pre-allocating a batch or testing a system before go-live.
What exactly does a PO number identify?
It identifies a specific commitment to buy, not the goods themselves and not the payment. Once a requisition is approved internally, the purchasing system issues a PO number and sends the order to the supplier. That number then appears on the supplier's shipping documents, the warehouse's goods receipt note, and finally the invoice the supplier submits for payment. If any of those documents shows a PO number that does not exist in the system, or one that has already been fully invoiced, accounts payable holds the payment until someone resolves the mismatch.
Should PO numbers be sequential?
In most organisations, yes, and this is not just a stylistic preference. Internal and external auditors routinely check for gaps in a PO sequence, because a missing number can mean a cancelled order that was never properly closed out, or in less benign cases, an order that was deliberately kept off the books. A clean, unbroken sequence is one of the simplest controls a finance team has for spotting irregularities, which is why most ERP systems auto-increment PO numbers by default rather than leaving the format open-ended.
That said, random numbers still have a legitimate place. Testing a new procurement system, populating a sandbox environment, or pre-generating a pool of numbers that a manager assigns in order once approved, all of these are cases where uniqueness matters more than strict sequencing. Our sequential code generator handles the production case with an incrementing {n} token, while the random generator suits testing and pre-allocation.
What format should the PO number actually follow?
A widely used pattern is prefix, year, running number: PO-2026-04831. The prefix makes the document type unambiguous the moment someone glances at it, which matters when a single filing system holds POs, invoices, and contract numbers side by side. The year segment speeds up archiving and year-end reconciliation. The running number does the actual job of guaranteeing that no two orders share an identifier. Some companies add a two- or three-letter department or cost-centre code as well, for example PO-MKT-2026-0483, which lets a manager filter their own department's orders without querying the full database.
Field length is a practical constraint worth checking early. SAP, Oracle, and Microsoft Dynamics all cap purchase order fields somewhere in the 10 to 20 character range, and a format that looks clean on paper can silently fail to save if it runs a few characters over the limit your specific ERP enforces.
How does a PO number connect to the eventual invoice?
When the supplier submits an invoice, most accounts payable teams run what is called a three-way match: they compare the PO, the goods receipt confirming what actually arrived, and the invoice itself. All three need to reference the same PO number and agree closely enough on quantity and price for the payment to release automatically. A separate invoice number gets created by the supplier at billing time, so the two numbers stay linked throughout the process but are never the same value, and a single PO can end up matched against several invoices if the order is delivered and billed in stages.
What goes wrong when PO numbering is not centrally controlled?
The most common failure is duplication across departments, where two teams each keep their own counter and eventually both land on, say, PO-1042. Accounts payable then either rejects the second invoice outright as a duplicate or, worse, matches it against the wrong order and pays the wrong line item. Centralising number generation, whether through the ERP's own sequence or a shared pre-allocated pool, removes this risk entirely once a company grows past the point where a single spreadsheet can track every order.
Frequently Asked Questions
What is a purchase order number used for?
It is the identifier that ties a specific request to buy goods or services to the eventual invoice and payment. Every document in the chain, the requisition, the PO itself, the goods receipt, and the supplier invoice, references the same PO number so accounts payable can match them up before releasing payment.
Should purchase order numbers be sequential or random?
Most accounting and ERP systems assume sequential PO numbers because gaps in a sequence get flagged during audits as a sign of missing or cancelled orders. Random numbers work fine for testing, sandbox environments, or pre-generating a batch that finance later assigns in order, but production systems should use a sequential generator for the real numbering.
What format do PO numbers usually follow?
A common pattern is a short prefix (PO or PUR), a four-digit year, and a running number, for example PO-2026-04831. The prefix makes the document type obvious at a glance, the year makes filing and archiving easier, and the running number guarantees uniqueness within that year.
How long can a PO number be before it breaks an ERP field?
Most major ERP systems, including SAP, Oracle, and Microsoft Dynamics, cap purchase order number fields somewhere between 10 and 20 characters. Check your specific system before settling on a format, since a number that is a few characters too long simply will not save.
What happens if two departments accidentally reuse the same PO number?
Accounts payable will either reject the second invoice as a duplicate or, worse, match it to the wrong order and pay the wrong supplier. This is exactly why centralised, unique numbering (rather than each department inventing its own counter) matters once a company has more than one person cutting purchase orders.
Is a PO number the same as an invoice number?
No. The PO number is created by the buyer when the order is placed, before any goods or services are delivered. The invoice number is created by the seller when they bill for it. A single PO can be invoiced multiple times if the order is split into partial deliveries, so the two numbers stay linked but distinct throughout the transaction.
Can I pre-generate a block of PO numbers for later use?
Yes. Generate a batch, export it to CSV, and let your purchasing team draw from that pre-approved pool as orders come in. This is common in organisations that require a finance sign-off step before a PO number is officially allocated to a specific purchase.