The purchasing process: six steps from request to payment

Everything in an effective purchasing process that happens before the purchase order is a decision. Everything after it is a commitment.

That distinction is worth holding onto, because it explains why the purchasing process needs controls that sourcing doesn't. A supplier shortlist can be revised. A negotiation can be reopened. But once a purchase order reaches a supplier, your organization has told an outside party to do something, and unwinding that costs money, goodwill, or both.

Most purchasing problems cluster around that moment. An order goes out before anyone confirmed the budget had room. A delivery gets signed for without being counted, and the shortage surfaces a month later when the full invoice arrives. An invoice gets paid because it looks legitimate, not because anyone checked it against what was ordered.

None of that is carelessness. It's what happens when the process lives in people's heads rather than in a defined sequence with a clear owner at each stage.

This guide walks through the six purchasing process steps, from the initial request to the paid invoice. For each one, you'll see what it produces, what needs to be settled before the purchase can proceed, and where it commonly stalls. After that, you'll learn the best practices that tighten the process without slowing it down, what to measure once it's running, and how purchasing fits inside the entire procurement cycle.

What is the purchasing process?

The purchasing process is the sequence an organization follows to buy the goods and services it has already decided it needs. The purchasing process begins with a formal request and ends with a verified, recorded payment.

Businesses buy differently from individuals, and the difference is authorization. A person decides and pays. An organization must show that someone with authority signed off, that funds existed, that what arrived matched what was requested, and that the bill reflects all of that. That verification process is the purchasing process. It exists because the person who wants something, the person allowed to approve it, and the person who releases the money are rarely the same.

Purchasing is sometimes described as transactional procurement. That's fair. It's the half of buying where documents get raised and money moves, as opposed to strategic sourcing, which decides what the organization ought to be buying and from whom.

Six stages cover the purchasing cycle in most organizations. What differs between companies isn't the sequence, but how much a system enforces and how much relies on someone’s memory. Purchasing teams buying raw materials for a production line and purchasing teams buying professional services run the same stages. They just weight them differently.

The six stages of the purchasing process

Each stage below closes off options the previous one left open. That's the useful way to read them. By the time an invoice is paid, roughly a dozen decisions have hardened into facts, and the cost of changing your mind rises at every stage.

For each one, the question that matters most isn't what happens, but what gets harder to undo.

Stage 1: Raise and approve the purchase request

A purchase requisition asks your own organization for permission to spend. It records what's wanted, how many, by when, at roughly what cost, and which budget line it comes from. It then enters the approval process, traveling to whoever holds spending authority at that amount, which in most companies is tiered so small buys clear locally and large ones climb.

The specification written here is the one thing from this stage that travels. Every document downstream inherits it. A quantity typed wrong in the requisition becomes a wrong purchase order, then a wrong delivery, then an invoice that matches the order perfectly and still isn't what anyone wanted. 

What gets harder to undo here: almost nothing, which is exactly why this stage deserves attention. It's the last point where changing the plan costs a conversation rather than money.

Where it stalls: requests waiting on approvers who are away, distracted, or unaware anything needs them. Because nobody can see who's holding a request, nobody chases it, and the delay only becomes visible when the requester gives up waiting and escalates. 

A purchase requisition system helps mainly by making both problems structural, not personal. Required fields mean approvers get complete requests, and a visible queue means the person waiting knows exactly where their request sits. A purchasing workflow in a system, not a habit, also means no purchase starts without someone accountable for it.

Stage 2: Research and compare suppliers

With permission granted, the question becomes who to buy from. A repeat purchase from one of your preferred suppliers skips most of this. A new category, a new supplier, or a large amount brings supplier research, a request for quotation, and a real comparison of potential suppliers.

What teams underestimate is how permanent a "let's try them once" supplier turns out to be. Bringing a new vendor on means banking details, tax documentation, insurance certificates, contact records, catalog data, and sometimes an integration. All of that is work, and none of it is work anyone wants to repeat. So the trial suppliers become pre-approved suppliers, often by default, on terms that were only ever meant to cover one order.

That's worth knowing before the first order rather than after the third. If a supplier is genuinely a trial, say so to them and set a review date. 

If they're likely to become the default, run the supplier evaluation you would run for a long-term partner. That means weighing supplier reliability, quality history, financial stability, and the administrative load they'll add against competitive pricing rather than after it. Evaluating potential suppliers on headline price alone is how an organization ends up with a reliable supply chain on paper and a fragile one in practice.

What gets harder to undo here: the supplier relationship, through accumulated setup rather than through any contract. Switching gets more expensive every month you don't.

Where it stalls: waiting for quotes. Potential vendors answer on their own timetable, and a request with no stated reply-by date runs at the speed of the slowest respondent.

Stage 3: Negotiate and agree terms

Price is the obvious subject of any contract negotiation, and often the least movable, particularly for a small buyer approaching a large supplier. Getting a reasonable price matters, but the terms around price are where the room usually is: when you pay, what counts as on time, what happens when something arrives broken, and how long the quoted price holds. 

Early payment discounts belong in the same conversation, because 2% for settling in ten days is often worth more than the price reduction you were arguing about.

The thing to understand about this stage is that whatever you agree now sets the ceiling, not the floor. Once an organization has paid a price for a year, asking for a lower one requires a reason that didn't exist at signature. Suppliers reasonably read continued payment as acceptance. The same applies to payment timing, which tends to get entered into the accounting system once and then treated as a fact about the supplier rather than a decision you made.

The corollary matters more: what you don't agree becomes what you can't claim. A remedy for late delivery is only ever needed on the day it's missing, and that's the worst possible day to start negotiating one.

None of this is adversarial, and it shouldn't be treated as such. Positive vendor relationships are built on terms both sides can live with for years, because a supplier who agrees to something unworkable will eventually find a way out. 

What gets harder to undo here: the pricing baseline and the payment terms, both of which become defaults through repetition rather than through any renewal clause.

Where it stalls: signing authority. Terms get settled, then someone discovers finance owns payment terms or the commitment period exceeds the negotiator's limit, and the whole thing rejoins a queue. Knowing your own limits before the call avoids nearly all of it.

Stage 4: Issue the purchase order

The purchase order (PO) turns an approved request and agreed terms into an instruction. It states the item, quantity, price, delivery date, and applicable terms, and carries a number that every subsequent document points back to.

This is the line the whole article is about. Up to now, everything has been internal. A purchase order sends an obligation outward. 

In many circumstances, an order the supplier accepts becomes a legally binding document, though whether it does and on whose terms depends on your jurisdiction and what's attached to it, which is worth confirming with your own advisors rather than assuming.

Here’s a concrete example. A facilities manager requests 40 replacement light fittings at an estimated $180 each, roughly $7,200, against a maintenance budget with $11,000 remaining. 

Approval clears the same afternoon. The supplier quotes $172 and a four-week lead time, so the order goes out at $6,880. At that moment, the maintenance budget still reads $11,000 in the accounting system, because nothing has been invoiced, but only $4,120 is actually available. 

Two weeks later, another request for $6,000 of the same budget looks affordable and is approved. Neither approver did anything wrong. The budget figure they both checked was describing the past. 

That is the general case. A department reads as comfortably underspent while sitting on issued orders nobody has invoiced yet. The money is spoken for; it simply hasn't moved.

Teams that can see committed and actual figures side by side get a far more honest answer to "how much is left," and carry noticeably less financial risk into month-end.

There's a second line most buyers never think about. For anything made to order, configured, or custom, the real point of no return isn't when you send the order. It's when the supplier starts work, which happens on their schedule and without telling you. 

Canceling before that costs a phone call. Canceling after it costs whatever they've already committed. Asking a supplier when that moment arrives is a short conversation that occasionally saves a great deal.

What gets harder to undo here: everything. This is the stage the rest of the process is arranged around.

Where it stalls: retyping. When purchase order creation means rekeying the request by hand, errors enter here and surface at invoice matching weeks later, long after anyone remembers the context.

The mechanics of creating, numbering, and structuring orders deserve more space than this article gives them. For that, see our guide to the purchase order process.

Stage 5: Receive and check the goods or services

When the delivery lands, somebody compares it against the order: count, specification, condition, completeness. For services, the equivalent is confirming the work met whatever standard the agreement set out.

Here's what makes this stage more consequential than it looks. Your right to reject usually has a deadline, set in terms almost nobody reads, and a signature at the door can count as acceptance. Inside that window, a short or damaged delivery is a supplier obligation. Outside it, the same problem becomes a favor. 

A second deadline runs alongside it and ignores the paperwork entirely. Once material enters production, gets installed, or is consumed, you can't return it, no matter what the terms say.

Whoever receives deliveries is frequently nobody from purchasing, and that's fine as long as two things are true: 

1. They know what was ordered
   
   
2. They have somewhere to write down what actually turned up. 

In many organizations, neither is reliably true, which is how a genuine shortage becomes an accounts payable mystery a month later.

Incomplete deliveries deserve specific care. Logging what arrived and what's outstanding keeps the order live and the gap visible. Closing it out because most of the shipment showed up hands the problem to someone with less information than you have.

What gets harder to undo here: your right to refuse. It expires on a clock you didn't set and probably haven't read.

Where it stalls: the record rather than the inspection. Telling a colleague the delivery looked fine isn't a receiving record. Delivery receipts that reference the order number are what let accounts payable close an invoice later without a phone call. When stock is involved, linking the receipt to inventory management also keeps the numbers honest for the next request.

Stage 6: Match, approve, and pay the invoice

The supplier invoice arrives, and the accounting department puts it against two other records: the order and the receipt. The three-way matching process asks a single question. Are we being billed for what we asked for and actually got, at the agreed price?

The recurring mismatches are unsurprising. Price differs from the order. Quantity billed exceeds quantity received. Or an invoice turns up with no order behind it at all, which normally means a purchase happened outside the process, also known as maverick spending.

Small differences shouldn't halt invoice approval, and most organizations set a variance allowance so minor gaps clear automatically and larger ones route to a human. 

Both extremes cost you: 

• Set it tight and the team spends its week on pennies
  
  
• Set it loose and real overbilling sails through.

Then comes payment processing, the final step and the least reversible part of the entire process. You can correct every earlier mistake internally, at some cost. You can recover an overpayment only if the supplier agrees, and a duplicate payment discovered after the fact becomes a collections exercise against a company you still need to buy from. This is the last checkpoint, and it's the only one where getting it wrong puts the remedy outside your control.

Paying within the agreed window is a real choice, not an administrative reflex. Early payment discounts are worth it when cash allows. Timely payment, made when you said you would, is one of the easiest ways to stay a customer a supplier prioritizes. Paying late without warning spends relationship credit that took effort to build. Paying everything the moment it lands surrenders flexibility for no return.

What gets harder to undo here: the money. After this, correction depends on someone else's cooperation.

Where it stalls: the missing receipt. Order present, invoice present, no record of what arrived. Accounts payable then goes looking for a confirmation that should have taken seconds at the point of delivery.

Three-way invoice matching, done automatically, compares the three records the moment an invoice is captured and sends only the mismatches to a person, turning invoice processing from a queue into an exception list.

Is the purchasing process the same as the procurement process?

No, but they aren't separate processes either. Purchasing is the transactional part of the overall procurement process, covering the six stages above.

The procurement cycle is wider at both ends. Before purchasing starts, the organization decides what it needs and whether it should meet that need by buying at all. After payment clears, supplier management takes over, reviews supplier performance, and feeds that into the next decision. 

Put simply, the procurement process includes everything purchasing does, plus two things purchasing doesn't. Mapped against the nine-step procurement process, purchasing runs from stage three to stage eight.

The gap between them shows up in what a team gets judged on. Purchasing teams answer for speed, accuracy, and price paid. Procurement teams additionally answer for whether the organization should have bought the thing at all, and whether the supplier will still be able to deliver in three years. 

The procurement process focuses on the decision. Purchasing activities focus on executing it without error. Sourcing belongs to procurement too, and it's genuinely distinct from purchasing: sourcing ends the moment you know who you're buying from, and purchasing is everything that follows. 

On one-off buys, the two typically collapse into a single event, which is part of why irregular spend drifts. Supplier selection ends up happening transaction by transaction, at precisely the moment when speed matters most and nobody is looking.

In smaller organizations, there is no separate procurement department, and one purchasing manager owns both halves. That works until spend grows past the point where a single person can hold the whole supplier ecosystem in their head, which usually happens well before anyone notices.

Where purchasing commitment outruns purchasing control

Most purchasing failures share a shape. A commitment gets made before the check that should have preceded it. Sorted that way, the failure modes are easier to fix, because each one names a missing control rather than a broken artifact.

At Tradogram, we have helped thousands of organizations move purchasing out of email and spreadsheets, and the same four failures come up in nearly every implementation.

An order goes out before the budget is confirmed. The approval covered whether to buy. Nobody verified the money was still there, because committed orders from earlier in the month weren't visible.

A purchase happens with no order at all. Maverick spend, usually on a card, usually because the sanctioned route felt slower than the workaround. Unauthorized purchases of this kind still commit the organization. It simply wasn't present for the decision. Treat a rising rate of this as a friction reading rather than a discipline problem, because that's almost always what it is.

A delivery is accepted rather than checked. Someone signs, the rejection window quietly closes, and a shortage becomes a negotiation.

An invoice is paid because it looked right. Not matched against anything, just plausible. This is the failure that can't be corrected internally.

Human error threads through all four, and it's worth being exact about what that means. These aren't random slips. They're the predictable output of manual re-entry, unclear ownership, and time pressure. Any process that requires nobody to ever be distracted will generate incorrect orders on a schedule.

Put the purchasing controls where the commitment is

The usual objection to tightening purchasing is that controls slow buying down and end up ignored. It's a fair concern. The resolution is that purchasing loses far more time to waiting than to checking, so the changes that streamline operations are the same ones that tighten control.

Three best practices do most of the work, and all three cluster around the order rather than spreading evenly across the process.

Set thresholds by value, and name a stand-in for every approver. Ask what each role can clear alone, then ask who covers when that person is out. Organizations get the first question right and forget the second, which is where a large share of stalled requests actually sit. Approval workflows should be loose enough that routine buys don't reach management and tight enough that significant commitments get read. If three signatures are normal, they're set wrong.

Write down when an order is required. Every purchase, or everything above a stated value. Either rule works. No rule doesn't, because accounts payable then has no way to know whether a matching order should exist. If you enforce an order-or-no-payment policy, define the exceptions in advance rather than arguing them case by case. This one rule does more for ensuring compliance than any amount of reminding.

Keep a list of suppliers already cleared to sell to you. For repeat and routine buying, a list of preferred suppliers with agreed pricing removes the slowest stage from most of your orders. It lets routine spending run cost-effectively without a fresh comparison every time, and it removes the main reason people go around the process.

That's deliberately three items rather than a full program. For the complete sequence, including documenting the process and consolidating your supplier base, the procurement process guide covers it in order of return.

The one number to track first

Measure purchase order cycle time before anything else: how long from a request being raised to an order reaching the supplier.

Use the median rather than the mean, because a handful of unusual purchases will pull an average away from what routine requests actually experience. Then break the total into three parts: time spent waiting for approval, time spent choosing a supplier, and time spent producing the order.

That split is the whole point. In our experience, approval waiting dominates in most organizations that measure it for the first time, and that finding changes what you fix. 

A team that hasn't segmented the number tends to attack the wrong segment, usually by pushing purchasing staff to work faster on the part of the process that was never slow.

Once cycle time improves and the process is followed consistently, broader measures of purchasing effectiveness become meaningful. Before that, they mostly describe variation.

The second measure worth adding is cost per purchase order, which tells you what your process costs to run rather than what you are buying. APQC's cross-industry benchmarking puts this between $14 and more than $54 per order, with the spread driven almost entirely by how much of the process is manual. Calculate your own by dividing the fully loaded cost of the purchasing function by the number of orders issued, then compare the two ends of that range against how many times a single order gets retyped in your process. 

The next most useful move is to analyze spending patterns by category and by supplier. That is where savings opportunities are visible without renegotiating anything: three departments buying the same item from three suppliers at three prices is a consolidation you can act on this quarter to reduce costs. Spending patterns only become readable once purchases are coded against categories consistently, which is why this follows the process work rather than leading it. Other procurement KPIs and how to read them are covered in the procurement process guide.

What procurement software changes about the commitment point

Nothing in this article requires software. Thresholds, better request forms, an order rule, a cleared supplier list: all of it works on paper, and organizations that sort those out first get considerably more from a system afterward.

What a system changes is narrower than most vendor material suggests, and for purchasing specifically it comes down to two things happening at the commitment point.

Requests reach the right approver without anyone routing them. Approval workflow automation applies your rules by amount, department, location, or category, identically every time. That's simultaneously a control and an accelerator, which is the part people find counterintuitive. A request can't skip an approver, and it also can't sit unnoticed, because the person waiting can see where it is. Since approval waiting is where most purchasing delays happen, that is where most improvement comes from.

Committed spend becomes visible before the invoice does. Budget and spend control shows what's already been promised alongside what's already been paid. That's the specific gap behind orders going out against budget that isn't there, and it closes at the moment an order is issued rather than weeks later when the bill arrives.

The rest follows from those two. Purchasing process automation carries information from request to order without retyping, removing errors that would otherwise surface at matching. Records stay linked, which is what makes cycle time and spend by category readable from work the team already did. 

Operational efficiency in purchasing is mostly the byproduct of never doing the same work twice, and what it buys back is room for strategic tasks: supplier evaluation, category planning, and the negotiations that actually move price. 

This is the connective tissue a procure-to-pay system provides, and it's why purchasing and accounts payable stop reconstructing each other's history.

Technical America, a manufacturer running purchasing across multiple sites, cut purchase order processing from two to eight hours to under two hours and reported roughly 25% purchasing cost savings after moving requests, approvals, and orders into one system. Read the full case study 

What decides whether that holds is fit. A platform that imposes one identical route on every department, site, or entity gets bypassed, and a bypassed process delivers neither speed nor control. Tradogram is built to be configured around your thresholds, categories, and structure instead.

If running the process consistently has become your constraint rather than deciding what to buy, see how the core features connect requests, approvals, orders, receiving, and invoices, or book a demo.