Procurement: Is There Another Source?

By nine on Tuesday morning Hannah Wolcott has the worst kind of instruction on her desk: find out whether there is a way out of this, and have an answer by Thursday.

It is the worst kind because everyone already assumes the answer is yes — that somewhere out there is another factory making control boards, and her job is to find it. The value she adds is not the source she finds. It is the three doors she closes firmly enough that nobody keeps hoping through them.

What lands on Hannah’s desk

Three things arrive together. The CB-40 specification: a board carrying the oven’s temperature and humidity logic, its safety interlocks and its firmware, wired a particular way inside three oven bodies. Module one’s arithmetic — 520 boards, 80 a week, 6.5 weeks of cover against a twelve-week delay, 440 ovens unbuilt.

And Ravi’s objective, binding on everyone downstream: protect the relationships that cannot be bought back at any price, recover revenue second, spend money only where it buys something lasting beyond the quarter. That last clause turns one of her four options from a waste of $180,000 into the most important thing she does all year.

She controls nothing about the design — that is Dev’s question, in module four. Hers is narrower: given this board, what can be bought, from whom, at what price, arriving when?

What a procurement manager actually does

In a normal quarter Hannah negotiates prices, manages contracts and keeps parts flowing into a factory that never thinks about her. A shortage turns her into the company’s professional sceptic about supply — the person who asks what evidence sits behind every claim made by a supplier, a broker or a hopeful colleague. A plan is only as good as the dates inside it, and Tomasz, Grace and Marcus all build on numbers she supplies. Report an optimistic date to please the room and you have poisoned six people’s work.

The software on Hannah’s desk

The supplier portal is where the delay first became visible, and it is worth noting that portals report status rather than intent: it showed a changed date, not the allocation decision behind it. That is why Hannah’s first move is a phone call — no system tells you where you sit in someone else’s priority queue.

Ariba or a similar sourcing tool carries the requests to Kestrel and the brokers, and more importantly carries the approvals, because buying $44,000 of parts from a broker is exactly the sort of decision that gets examined later. The comparison happens in Excel, as it always does. And the purchase order raised in the ERP is the moment an option becomes a commitment — which is why Ravi’s rule about no broker orders on day one was a rule about this screen.

The decisions

Getting the truth out of Vantor

“Twelve weeks” is not a fact but an opening position — a number chosen in Penang with enough padding in it that nobody has to send a second email.

So Hannah books a call for eight on Wednesday morning, four in the afternoon in Penang, and goes after five things. Is the delay allocation or capacity — has the chip maker cut what Vantor receives, or has Vantor’s own line stopped? The first is a queue, and queues can be joined at a different position; the second is physics. Where does Calder sit in Vantor’s queue, and what would move it up: volume, a longer contract, prepayment? Can they part-ship, since thirty boards in week seven is thirty ovens out of the door? And what happens in week thirteen, when at eighty a week Calder never catches up its backlog?

Her leverage is poor: Calder buys four thousand boards a year, which Vantor’s plant makes in a fortnight. Small customers rarely get priority — but if they behave well they get truth, because there is no commercial reason to manage them. A believable date is what she is really shopping for, since a plan built on an unreliable one fails later, when the alternatives have expired.

There is a version of this call that feels better and works worse. Vantor has just cost Calder $5.3m and the urge to say so is real, but boards are rationed inside Vantor by people with discretion, and discretion goes to whoever is straightforward to deal with. Shout, and your supplier stops telling you anything before it is certain — exactly when you need to know. Hannah asks for written confirmation every Friday, but without theatre: her most valuable asset for twelve weeks is a person in Penang who picks up the phone.

The alternative supplier, and why ten weeks

You cannot simply order the CB-40 from someone else, and the reason is not procurement bureaucracy.

Kestrel Micro in Poland can build a board that does what the CB-40 does — but that must be proven five ways before one oven may carry it. Electrically identical: same voltages, same signals to the heating elements and the gas valve, same behaviour when a sensor fails. Physically identical: same footprint, mounting holes and connectors, because the loom and the sheet metal were designed around the original. The firmware ported and validated, because a control loop that overshoots by four degrees is invisible on a bench and ruins a bakery’s product for a year. The factory audited: Calder is buying a repeatable process, not a design. And samples built and tested in real ovens.

Then the immovable one. A finished oven carries UL listing, electrical and gas safety approval, and those approvals reference the certified design, including its control board. Change the board and you are altering an approved safety-critical assembly: test-house work and documentation before anything can legally be sold.

Add it up: ten weeks, $180,000 of tooling, boards at $460 against Vantor’s $320 — forty-four per cent more, permanently. Dual sourcing is not free and never was; the saving that created this crisis was somebody, years ago, quite reasonably declining to pay it.

The gap opens in week seven; Kestrel’s first approved boards land in week eleven at the earliest, and only if nothing slips. Kestrel does not rescue this crisis. It insures the next one.

Hannah starts it on Wednesday anyway, because of Ravi’s third clause: this is the only money on the table that buys something outliving the quarter. A crisis is also the only time it gets approved — ask in a calm quarter and you are told to cut costs; ask in week one of a shortage and it is signed by Thursday.

The broker temptation

By Tuesday afternoon three brokers have emailed unprompted, offering CB-40 boards for immediate delivery at around $1,100 — three and a half times list. Hannah contacted none of them.

Brokers hold stock from cancelled orders, from distributors clearing an overhang, from firms that went under mid-build, and from sources nobody will name. Some is genuine and boxed. Some is not: counterfeit and recycled parts are a documented problem — components pulled off scrapped assemblies, re-marked as a higher grade, sold with unverifiable paperwork.

The stakes are not a spreadsheet variance. A substandard board sits inside a commercial oven in a hotel kitchen, governing a gas valve, a heating element and the temperature at which food is cooked. The exposure is fire, food safety and a recall across three lines — and the liability does not stay with the broker. Calder’s name is on the oven.

A blanket ban is the safe answer; three hundred boards bought on Thursday because somebody senior demands action is the dangerous one. Hannah takes the middle path: one broker with a real trading history, full chain of custody back to an authorised distributor, and forty boards for about $44,000, each tested individually and reserved for the highest-value builds — because a $780 premium is a rounding error against a Rack oven’s $6,820 of gross profit per board, and half the margin against a Compact’s $1,560.

And they are not counted in the 520. Until they pass test they are a hope, not stock, and Tomasz plans against what exists. A board that might be counterfeit is not a board.

Expedite, and what $85,000 actually buys

When Vantor resumes, the batches can be flown rather than shipped: $85,000, and about two weeks clawed back. Simple arithmetic, subtle judgement.

Two weeks of production is 160 ovens — roughly $1.9m of revenue at an average price near $12,000 and, at the blended 33% margin, about $630,000 of gross profit. Against $85,000 that is not close — and the lopsidedness is what makes it dangerous, because it holds only if the boards arrive when promised.

Air freight is booked and paid against a release date. If Vantor slips again the aircraft still saves two weeks against sea, but in week fourteen rather than week eleven — after Bellwether’s dry dock window has closed and Nordfoods has moved its volume. Two weeks of the wrong ovens, arriving after the customers they were meant to save have gone.

So this is not a cost decision but a confidence decision, and Hannah frames it that way because Claire prices it in module eight: $85,000 against $630,000, multiplied by the probability that Vantor’s date holds. What she owes Claire is not a recommendation but an honest number for that probability, and a trigger — nothing booked until Vantor confirms in writing.

Where this goes wrong

Four options go on one slide and within a week a line in a board pack — “we are qualifying a second source” — has become “Kestrel is coming” in six people’s heads. Nobody decided to believe it; the ten weeks simply stopped being attached to the ten-week thing. Then the sales director, hearing boards are on the way, holds off the difficult conversation with Nordfoods for a fortnight — and that fortnight was the only thing that would have saved the account.

So every option Hannah reports carries a date, a probability, and a sentence about what it does not solve. Kestrel: week eleven at the earliest, nothing in weeks seven to ten. Brokers: forty boards, not counted until tested.

What Hannah hands on

To Ravi and Tomasz goes the sentence the company needs by Thursday afternoon: no rescue arrives inside the window. The 520 boards in Rockford are all there will be, so this crisis will be survived by allocation and customer management, not by supply. Tomasz’s job in module five exists because of that sentence.

To Dev goes the question she cannot answer: since swapping suppliers takes ten weeks because the design is fixed, could a different component be designed in, and what would that cost in recertification? She hands over what Vantor has told her about which microcontrollers can actually be bought.

To Claire goes the price of each lever with its conditions: $85,000 for two weeks, subject to a confirmed date; $180,000 of tooling plus $140 a board, for a benefit landing after this quarter; $44,000 of broker stock that may yet be written off. Finance does not need her recommendation. It needs the shape of each bet.

Spot the decision

Read each situation and decide how a procurement manager should handle it, then tap a card to check.