PO vs Non-PO Invoices: Handling Both in Construction AP
Every invoice that arrives in a construction AP department takes one of two paths. Either it was preceded by a purchase order — someone authorized the spend, in a system, before the goods or services arrived — or it was not. That single fork determines how the invoice can be validated, who has to approve it, how much control sits behind it, and how much manual work it takes to process. It is the most important distinction in AP, and construction's particular spend profile makes it especially consequential.
The trouble is that contractors generate an unusually large share of non-PO spend, and non-PO invoices are exactly where the control gaps and the manual effort concentrate. A PO invoice can be matched against the order; a non-PO invoice cannot, so it leans on human judgment instead. This post lays out the real difference between the two, explains why construction produces so many non-PO invoices, names the control gap that creates, describes how to route, code, and approve each path well, covers two-way versus three-way matching by path, and explains how to pull recurring non-PO spend back under control with blanket POs — and what PO coverage rate to actually aim for.
A PO-backed invoice has a paper trail that begins before the spend. A purchase order is created, the spend is authorized at that moment, and the order is sent to the vendor. When the invoice later arrives, AP has something concrete to check it against — agreed prices, agreed quantities, an authorization that already happened. The control is front-loaded: the decision to spend was made, and recorded, before the money was committed.
A non-PO invoice has no such trail. The first time the organization sees the spend in a system is when the bill arrives. There is nothing to match against — no agreed price, no agreed quantity, no prior authorization. The control, if there is any, has to happen entirely after the fact: someone looks at the invoice and decides whether it is legitimate and whether the company should pay it. The difference matters because front-loaded control is structurally stronger than after-the-fact control. With a PO, an unauthorized or mispriced commitment can be caught before it becomes an obligation. With a non-PO invoice, the obligation already exists by the time anyone looks.
A blunt way to put it: a purchase order moves the spending decision to before the money is committed. A non-PO invoice leaves the spending decision until after the company already owes it. Everything else about handling the two paths follows from that.
In a manufacturing environment, nearly all spend runs through purchase orders, and PO coverage above 90% is normal. Construction is structurally different, and a large share of its spend is non-PO not through sloppiness but through the nature of the work.
The construction spend that naturally arrives without a purchase order
- Subcontractor pay applications — billed against a contract and a schedule of values, validated by percentage of work in place, not against a PO
- Field-initiated purchases — small tools, fasteners, and supplies bought from a supply house mid-task to keep a crew moving
- Equipment rentals — extended week by week as the schedule shifts, with the bill arriving before anyone formalized the extension
- Utilities and temporary services — power, water, fuel, and portable facilities billed on a recurring basis
- Professional and indirect services — design, engineering, testing, and inspection invoiced against an engagement rather than an order
- Change-driven spend — work that arose from a field condition or a change directive and was bought before procurement caught up
The thread connecting these is urgency and dispersion. Construction spend is initiated by people in the field, on the day, under schedule pressure, away from a procurement system. A crew that needs a part to finish a pour today is not going to wait for a PO to be raised. The result is that a contractor's non-PO share is often a large minority — or even the majority — of invoice volume. That is the reality to manage, not a behavior to simply scold out of existence.
Because a non-PO invoice has nothing to match against, it carries a specific and well-understood set of risks. These are the gaps a contractor's AP process has to close deliberately, because the structure of a non-PO invoice does not close them on its own.
What can go wrong on a non-PO invoice that a PO would have caught
- Unauthorized spend — a purchase no one with budget authority actually approved becomes an obligation before anyone reviews it
- Price creep — there is no agreed price, so an inflated or drifting rate has no order to be checked against
- Duplicate payment — without a PO to tie invoices to, a resubmitted or doubled invoice is harder to catch automatically
- Wrong cost coding — with no PO carrying the job and cost code, coding falls to whoever processes the invoice, and errors distort job-cost reports
- Maverick spend — purchases made outside negotiated supplier agreements, forfeiting volume pricing and contract terms
- Weak fraud defense — a fabricated invoice from a fake or ghost vendor has no order to contradict it, so it relies entirely on a reviewer noticing
~0x
Relative cost-to-process of a non-PO invoice versus a PO-matched invoice, driven by the manual coding and review it requires (IOFM)
Note the second cost. A non-PO invoice is not only riskier; it is more expensive to process, because the validation, coding, and approval that a PO would have automated all fall to a person. Reducing non-PO volume is therefore both a control improvement and an efficiency improvement — the same move pays off twice.
The goal is not to force every invoice down one path — that is impossible in construction. The goal is to handle each path with a process suited to it: lean and automated for PO invoices, controlled and deliberate for non-PO ones.
A PO-backed invoice should be the easy one. The job and cost code travel with the purchase order, so coding is inherited rather than decided. Approval largely happened when the PO was authorized, so a clean, in-tolerance match needs little or no further sign-off. The PO invoice is the natural candidate for straight-through, touchless processing — and the design objective is to keep that path frictionless so the AP team's attention is free for the invoices that need it.
A non-PO invoice needs the control that a PO would have provided, delivered after the fact. That means routing it to the person who can confirm the spend was legitimate — typically the project manager or the budget owner closest to it — and capturing correct job and cost coding as part of that step rather than guessing at it in AP. Pay applications are a distinct sub-case of the non-PO path: they are validated against the contract value, the schedule of values, and the percentage of work in place, and routed to the PM who can vouch for that percentage. The non-PO path should be efficient, but it cannot be unattended; a person has to stand behind the spend.
“Our mistake for years was running every invoice through one queue. Splitting PO from non-PO changed everything — the PO invoices basically process themselves now, and our reviewers spend their day on the non-PO spend, which is exactly where the actual risk was sitting.”
— AP Manager, mid-market general contractor
Get AP insights in your inbox
A short monthly roundup of construction AP + accounting posts. No spam, ever.
No spam. Unsubscribe anytime.
Two-Way Versus Three-Way Match by Path
Matching is the mechanical heart of validating a PO invoice, and the right kind of match depends on what the spend is.
A three-way match compares the invoice against the purchase order and the receiving record — price agrees with the order, quantity agrees with what was actually received. It is the strongest control and the right default for material purchases against a PO, because it confirms not only that the price was agreed but that the goods genuinely arrived. A two-way match compares the invoice against the PO alone, with no receipt. It suits PO-backed spend where a receiving record does not naturally exist — many services, for instance — and it still validates price and authorization even though it cannot confirm delivery.
A non-PO invoice can do neither. With no purchase order and no receiving record, there is nothing to match against — which is precisely why the non-PO path substitutes human review and budget-owner approval for the automated match. Pay applications, again, are their own model: they are validated against the contract, the schedule of values, and verified percentage complete rather than through any two- or three-way match. The practical rule: three-way match material POs, two-way match service POs that lack a receipt, and route non-PO invoices to a reviewer because matching is not available to them.
Three-way matching only works if receiving records actually get created. On a jobsite, that means delivery tickets get captured and tied to the PO. A three-way match policy with no reliable receiving step quietly collapses into a two-way match — make sure the receiving habit exists before relying on the control.
Not all non-PO spend has to stay non-PO. A large share of it is recurring and predictable, and recurring predictable spend is a strong candidate to be pulled back onto a purchase order — specifically a blanket PO.
A blanket PO is a single standing order covering repeated purchases from a vendor over a period, up to a total value, against which many invoices are drawn. It is well suited to construction's recurring non-PO categories: an equipment rental running for months, an ongoing supply-house account, a recurring service. Putting that spend on a blanket PO converts a stream of unmatched non-PO invoices into a stream of invoices that can be two-way matched against the blanket — restoring price and authorization control, fixing the coding once on the PO instead of repeatedly in AP, and giving each invoice an automated path it otherwise lacked.
Recurring non-PO spend worth converting to blanket POs
- Equipment rentals that run for weeks or months on a job
- Standing supply-house and material-yard accounts used repeatedly across jobs
- Recurring services — testing, inspection, waste hauling, temporary facilities
- Utilities and temporary services with a predictable monthly billing pattern
- Any vendor that invoices the same job repeatedly on a regular cadence
This is exactly the kind of pattern an AP automation platform can surface. Covinly, for example, identifies vendors invoicing the same job repeatedly without a PO — the recurring non-PO spend that is a candidate to move onto a blanket PO — and routes each path accordingly: PO invoices toward straight-through matching, non-PO invoices to the right reviewer with the job-cost coding already proposed. The platform does not eliminate non-PO spend, but it makes the recurring, convertible part of it visible so a contractor can act on it deliberately.
PO coverage — the share of invoices, or of dollars, backed by a purchase order — is the headline metric here, and the temptation is to chase the manufacturing benchmark of 90%-plus. For a construction company that is the wrong target, and chasing it leads to either frustration or fake POs raised after the fact, which provide control theater rather than control.
The realistic frame is to accept that a meaningful share of construction spend — pay applications above all, and genuinely urgent field purchases — will always and appropriately be non-PO, and to focus PO coverage on the spend that can be planned. Set the target by segment: push recurring and plannable spend toward high PO coverage through blanket POs, leave pay applications on their own contract-validated path, and accept a residual of true ad-hoc field spend. Measure coverage by dollars as well as by count, since the dollars are where the working-capital and control stakes sit. A contractor who moves a large fraction of recurring non-PO spend onto blanket POs has materially improved control and efficiency — even if the blended PO coverage rate never approaches a manufacturer's, and it should not be expected to.
PO and non-PO invoices are not a problem to be solved by forcing everything into one mold. They are two genuinely different kinds of spend, and construction will always produce a lot of both. The contractors who handle AP well do not pretend otherwise — they run two deliberate paths: a lean, matched, mostly touchless path for PO invoices, and a controlled, reviewer-backed, carefully coded path for non-PO invoices. They keep working the recurring non-PO spend onto blanket POs to shrink the harder path over time. And they set a PO coverage target that respects construction's reality. Two paths, each suited to its spend, is not a compromise. It is the design.
Written by
Sarah Blake
Head of Product
Former AP Manager at a $200M construction firm, now leads product at Covinly. Writes about what AP teams actually need from automation — beyond the marketing promises.
View all posts