When Do You Need a Purchase Order? PO Thresholds and Policy
Jul 11, 2026
Jul 11, 2026
Convert a purchase order to Excel, CSV, or JSON
Submit your purchase orders
Drop documents here, or click to file
Up to 50 files per batch
Uploading...
You need a purchase order when a purchase is large enough, recurring enough, or risky enough that you will later want to prove what was agreed. Most US businesses set a dollar threshold, commonly somewhere between 250 and 1,000 dollars, above which a PO is required, and let smaller routine spend run on a card or an expense claim. The test is simple: if you would ever need to dispute the invoice, verify what arrived, or show an auditor the approval, raise a PO.
Last updated July 2026.
Purchase orders get a bad reputation because too many companies apply them to everything, including the 40 dollar box of printer paper, and then wonder why people stop following the process. The point of a PO is control where control is worth having. Below is how to decide where that line sits, and what the alternatives are underneath it. The tool above turns the POs you do raise into clean line-item data, so the control does not come at the cost of hours of typing.
A PO does four things. It records what was agreed, in writing, before the money moves. It creates an approval point, so somebody with authority signs off before the commitment exists. It gives receiving something to check the delivery against. And it gives accounts payable something to match the invoice to, which is the whole basis of the three-way match.
If none of those four things matter for a given purchase, the PO is bureaucracy. If any of them do, it is cheap insurance. That is the entire decision.
| Situation | Raise a PO? | Why |
|---|---|---|
| Purchase above your dollar threshold | Yes | Approval and audit trail before commitment |
| Recurring supplier relationship | Yes | Locks price and terms, supports matching |
| Anything with a delivery to verify | Yes | Receiving needs something to check against |
| Custom, made-to-order, or long lead time | Yes | Disputes are likely and costly without one |
| Spend charged to a grant, job, or client | Yes | The cost must be traceable to its funding source |
| Small routine supplies under threshold | No | Card or expense claim is cheaper to administer |
| Utilities, rent, payroll, subscriptions | No | Governed by a contract already; a PO adds nothing |
| Genuine emergency purchase | Exception | Buy now, document under a defined exception policy |
The threshold is the dollar figure above which a PO becomes mandatory. There is no universal right answer, and copying another company's number is a mistake, because the correct threshold depends on how much other control you already have. Common practice in US small and mid-sized businesses lands somewhere between 250 and 1,000 dollars, and larger organizations often run higher because expense and card controls catch the small stuff.
Set it by working out what the process costs you. If raising, approving, and matching a PO consumes 20 minutes of staff time, applying it to a 50 dollar purchase is a net loss even before anyone gets annoyed enough to bypass it. Set the threshold high enough that the control is worth its administrative cost, then use cards, expense policy, and spot review to cover what falls below.
Watch for threshold splitting once you set the number. Two orders of 4,500 dollars where one of 9,000 belongs is the most common way a threshold gets defeated, and it is worth reporting on. It is a classic form of maverick spend, and it usually means your approval process is slow rather than that your people are dishonest.
Below the line, the goal is a light control that still leaves a record. Purchasing cards work well for routine, low-value, high-frequency spend, as long as the categories are defined and the statements are reviewed. Expense claims cover one-off buying by individuals. Standing agreements and blanket purchase orders handle repeated buying from the same supplier without raising a fresh PO every time, which is the right answer for consumables you order weekly.
Whatever you choose, decide it explicitly and write it down. Most companies do not have a policy gap so much as an undocumented one, where everyone follows a different rule they believe is correct.
When the purchase is above your dollar threshold, involves a supplier you buy from repeatedly, has a delivery someone must verify, or will be charged to a grant, job, or client that requires traceability. In short, whenever you might later need to prove what was agreed, check what arrived, or show an auditor who approved it. Small routine spend below the threshold does not need one.
No. There is no US law requiring a purchase order for a business purchase. It is a commercial and internal control document, not a statutory one. Once a supplier accepts it, though, a PO generally forms a binding contract for that order, which is exactly why it is useful: it fixes the price, quantity, and terms in writing before the goods ship.
US small and mid-sized businesses commonly require a PO somewhere above 250 to 1,000 dollars, with larger organizations setting higher limits because card and expense controls already cover smaller spend. The right number depends on how much the PO process costs to run. If administering it costs more than the risk it removes, the threshold is set too low.
For their larger and recurring purchases, yes. A small business gains the same things from a PO that a large one does: an approval point, an agreed price, and something to match the invoice against. The difference is scale, not principle. What a small business should avoid is applying the PO process to every minor purchase, which creates work without adding control.
Routine low-value supplies, utilities, rent, payroll, insurance premiums, and software subscriptions typically do not. Most are already governed by a contract or a fixed agreement, so a PO adds a step without adding control. Small incidental buying is usually better handled with a purchasing card or an expense claim, which cost far less to administer per transaction.
You lose the approval point and the matching basis. When the invoice arrives, accounts payable has nothing to check it against, so overbilling, duplicate invoices, and quantity disputes are much harder to catch. It also produces retroactive POs raised after the fact to satisfy the system, which turns approval into a formality and defeats the control entirely.
A PO policy fails for one of two reasons: it applies to too much, or the administrative burden of following it is too high. The first is fixed by setting a sensible threshold. The second is fixed by removing the typing, which is where most of the burden actually lives. Reading supplier POs and confirmations by hand into a spreadsheet or ERP is the step that makes people quietly stop raising them.
That is what this tool removes: upload the purchase order and get the vendor, PO number, dates, and every line back as clean rows in about ten seconds, ready for QuickBooks or whatever system you keep the books in. If you are formalizing purchasing policy generally, it is worth setting up vendor requirements at the same time, since most companies that require a PO from a service vendor also want proof of insurance on file, and tracking those certificates is a separate discipline that tends to get forgotten until a claim.
For the mechanics of the process itself, the purchase order process guide walks the full requisition-to-payment path, and purchase order extraction for accounts payable covers the matching end where the PO finally pays off.