A purchase order is the last document a large buyer produces and close to the only one a supplier is ever handed whole. It carries a number, a delivery address, a payment term nobody negotiated in the seller's hearing, and a line of description that came from somewhere else entirely. Everything that decided it happened earlier, inside the buying organization, in a sequence the order itself does not record.
So this article takes one order and reads it backward. Each section below is a hand the order passed through on the way to being signed, taken in reverse, and each one closes on the same pair of questions: what was still open when the requirement reached that hand, and what had already shut. Read forward, corporate purchasing looks like a decision about a supplier. Read backward, most of it turns out to be a decision about a category, taken months earlier by somebody who will never take a sales call.
The last hand, and the only one that signs
Somebody had to be permitted to commit the money. In most large organizations that permission is set by amount rather than by judgment. A manager can approve to one level, a director to a higher one, and above some figure the purchase stops being a decision at all and becomes a procedure with named steps. Those bands are internal policy, written to keep the organization's own staff from spending badly. Their effect on outside suppliers is a side effect nobody in the building spends any time thinking about.
The side effect is not small. A quoted price sitting just above a band is not slightly more expensive than one sitting just below it. It is a different process, with more approvals, more documentation and a longer wait for the person who wanted the thing in the first place. Staff who live inside that process learn where the lines fall and shape requirements to stay beneath them, which is one reason a supplier occasionally loses work it was genuinely cheapest on.
What the bands actually are, at any particular buyer, is not something this publication can report, and the reason is worth stating rather than stepping around. Purchasing policy at a private company is not published anywhere it could be read and cited. A good deal of the public equivalent is not published either. Dallas Fort Worth International Airport states on its procurement pages that it is a public entity and must follow applicable federal and state law when buying goods and services, and the dollar thresholds at which its own formal competitive bidding begins do not appear on those pages at all.
The hand before it: a requisition, and the list it was checked against
Before an order exists there is a requisition, raised by whoever actually needs the thing. The system it is raised in asks immediately for a vendor record, and a firm without one cannot be selected in that field. It is a narrow technical fact with a wide commercial consequence.
An approved vendor list gets described to sellers as a shortlist of firms the buyer rates. It is better understood as a register of firms the buyer has finished checking and is able to pay. Getting onto one means a tax identification number, insurance certificates carrying the buyer's own limits, banking details, agreed payment terms, and signatures on a code of conduct and whatever else the legal department attaches. Almost none of that measures capability, and much of it is handled by people the seller never speaks to.
Which explains why the list behaves as it does. The person raising a requisition is not resisting new suppliers out of loyalty to old ones. That person is choosing between a name that can be typed into a field this afternoon and a name that requires a form, an approval and a wait of unknown length. Being absent from the list is not the same as having been rejected. It means being expensive to choose, which produces the same result and teaches the supplier nothing.
The hand before that: whoever owns the category
Large buyers group their spend into categories and give each category an owner. The category, not the individual purchase, is the unit those people are measured on. A category owner is asked to consolidate suppliers, hold prices against last year, reduce how many firms carry critical risk, keep supply continuous through disruption, and report on all of it upward on a fixed calendar. Every one of those is a portfolio objective. A new supplier is therefore a change to a portfolio long before it is a purchase, and it competes for attention against whatever else is changing in that portfolio this quarter.
Job titles here are unreliable, and the unreliability is documented in public where anyone can check it. Dallas Fort Worth International Airport labels one function Business Diversity and Development in its site navigation and Business Development in the page text beneath, where the group is called the BDD Team. If a public buyer's own pages carry two names for one department, a private buyer's title printed on a business card is thin evidence about who decides anything.
This is also the hand carrying a reporting obligation that some suppliers are in a position to help satisfy. What that reporting actually counts, and what it cannot do for a firm that qualifies for it, is set out in what a diversity program actually counts.
The first hand: whoever wrote the specification
The specification is the earliest hand and the one with the most left to give away. It gets written by a technical lead or by the category owner, usually with something already in mind and often with an incumbent's product literature open alongside. Evaluation criteria are then derived from it, which is why the criteria so often read like a description of one particular firm. By the time anything becomes competitive, the competition is a contest over who best matches a description that already exists.
A supplier that has never spoken to whoever writes specifications is left competing on the two variables the description leaves free, which are price and delivery. Those are the two dimensions on which an entrenched incumbent is hardest to beat, and the two on which a buyer is least willing to accept an unfamiliar name.
The four hands, put back in order
Set the hands out again in the sequence the requirement actually met them and the reachable parts separate cleanly from the rest. The specification was open to influence and is closed by the time it is published. The category owner is reachable, though only between review cycles and only on the terms of the category rather than on the terms of a product. The vendor list is reachable at any time and is mostly an exercise in paperwork, which is exactly why it is the piece worth finishing before anything else is attempted. The signature is not reachable at all, and the sales effort aimed at it is the effort most often wasted.
That ordering feels wrong because the advice a small firm usually receives runs the other way, starting with a meeting and ending with forms. Forms first is duller. It is also what the machine was built to reward, and the machine is not going to be talked out of its own design by a supplier.
What cannot be told to a reader about any particular buyer
No large private buyer consulted for this article publishes its approval bands, its category structure or its supplier onboarding standard. Those documents exist. They are handed to staff, and in part to suppliers already onboarded. They are not public and cannot be cited, and a publication that prints figures from them is reporting a memory rather than a source.
So the shape gets described here and the numbers do not. The shape is the stable half anyway. It repeats across buyers in an industry and across industries in a way no threshold figure ever does. The public side of the same subject is far more legible by comparison, because a government buyer's authority to spend has to be written down where a citizen can find it, and the Comptroller's own bidders list page prints Texas Government Code 2155.263-270 as the authority it runs on. That asymmetry between what a public buyer must disclose and what a private buyer may keep is itself part of the subject of how buying organizations are put together.
A supplier that can name the hand it keeps losing to can stop paying for approaches to the other three. Working out which hand it is takes evidence rather than instinct, and evidence means bid tabulations, debriefs, onboarding checklists and the buyer's own published policies where any exist. The rest of the wider review this section sits inside is arranged around questions of that kind rather than around encouragement.