↓ Download as PDF

There is a moment in every scaling hardware company when the person running procurement realises the ground has shifted under them. For years the job was to take a defined bill of materials, put it out to bid, and drive the unit price down. Then demand takes off, the category tightens, and the supplier who used to chase the order stops returning the call — not because the relationship soured, but because they are now allocating scarce output and deciding who gets it. The buyer still thinks they are choosing the supplier. In a constrained market, it is the other way round. And the companies that keep behaving as though they are in charge are the ones that end up short of the one component that gates their whole schedule.

Sourcing is not procurement wearing a bigger title

Start by separating two activities that get filed under the same word. Project procurement takes a known design and a known schedule and buys it well: the right part, the right price, on time. It is a real discipline and it matters. But it optimises for the next delivery. Sourcing strategy does something different and earlier — it secures the right to buy at all, at the volumes you will need, in the years before those volumes exist on any purchase order. One is executed against a BOM. The other is executed against a forecast, a market structure, and a set of suppliers who have not yet agreed to grow with you. Run the second with the habits of the first and you will get a clean audit trail and an empty loading dock.

The reason this matters more in a growth company than a mature one is arithmetic. A stable business buys roughly what it bought last year; supply is a solved problem and price is the game. A company doubling is asking its supply base to double with it — and a supplier’s capacity does not double because you sent a larger order. It doubles when the supplier decides to build a line, and that decision is made on their timeline, with their capital, in response to a demand signal they choose to believe.

The constraint is belief, not capital

It is tempting to assume that in a shortage the buyer with the deepest pockets wins. Sometimes, briefly, at a spot premium, they do. But durable access to a constrained category is not bought at auction. Suppliers with more demand than capacity do not allocate to the highest bidder this quarter; they allocate to the customers they believe will still be placing orders when today’s project is finished — the ones whose demand is real, funded, and recurring. Allocation is a bet on your future, and the currency you pay it in is credibility, not cash.

That single fact rearranges the whole job. It means your forecast is not an internal planning artifact — it is a sales document you present to your own suppliers, and its credibility is a competitive asset you can spend or squander. It means the durable, recurring parts of your demand — the refresh cycles, the maintenance base, the second and third projects behind the first — are worth more to a supplier’s allocation decision than a single large launch order, and you should say so out loud. And it means a long-term agreement is not primarily a price instrument. It is a mechanism for converting your belief in your own pipeline into a supplier’s belief, early enough that they build capacity against it.

The corollary is the rule that sits underneath everything else here: you can accept concentration, but you can never accept being unwittingly single-sourced. Managed concentration — a deliberate, eyes-open decision to lean on one supplier because the alternative is worse — is a legitimate strategy. Discovering, the week a line goes down, that you had no second source and no plan for one is not a strategy. It is an accident you chose not to prevent.

Not every category deserves the same strategy

The next mistake is treating the supply base as one problem. It isn’t. In almost every hardware system, a small number of categories carry most of the money and most of the risk, and a long tail carries neither. Map your spend and you will find the familiar shape: a few categories dominate the cost of the finished system, and the rest, however numerous, are rounding error. Attention is your scarcest resource. Spend it where the Pareto tells you to.

But spend alone is the wrong sorting key, because a cheap part on a single-source, long-lead process can stop your line just as dead as an expensive one. The tool that has survived forty years of use because it is right is Peter Kraljic’s — classify every category on two axes at once: how much it matters to you (spend and criticality), and how exposed you are on the supply side (few sources, fragile process, long lead time). Those two axes make four quadrants, and each quadrant is a different job.

The value of the grid is not the labels; it is the discipline of admitting that the aggressive price posture you use in the Leverage box will lose you the Strategic supplier you were supposed to be partnering with, and that the partnership posture you extend to a Strategic supplier is money left on the table if you use it on a commodity. One strategy across all four quadrants is the single most common and most expensive sourcing error I see.

For the categories that matter: anchor, then deliberately second-source

Take the categories that land in the Strategic and Bottleneck boxes — the ones that decide whether you ship. The sequence that works is consistent, and the order is not negotiable. Secure share-of-capacity with an anchor supplier first. In a tight market, allocation comes before price; a lower quote you cannot get delivered is worth nothing. Lock a credible position with the supplier best able to carry you now, using the length and certainty of your demand as the lever.

Then, in parallel and without disrupting that anchor, develop a qualified second source. Not because you intend to split the business tomorrow — you may never split it — but because a qualified alternative is what converts you from captive to concentrated-by-choice. Second-sourcing is slow: qualification, first articles, process validation, all of it takes real time, which is precisely why it has to start long before you need the result. Frame it to the anchor honestly, as resiliency rather than a competitive threat, and structure it so the anchor keeps right of first refusal at matching terms. A good anchor understands supply-chain resilience; a supplier who insists you stay single-sourced to them is telling you something about how they will behave the day you are most exposed.

A sourcing strategy that lives in a deck is already decaying

None of this holds by itself. The map goes stale, the qualified second source quietly lapses because nobody placed the token orders that keep it warm, the forecast you gave your suppliers drifts from the one your board believes, and one day the whole carefully-built position has eroded without a single decision to erode it. The defence is governance, and it is unglamorous: a named owner for each strategic category who is accountable for its supply position, not just its price; a live picture of what is single-sourced, what is dual-qualified, and where each anchor’s allocation actually stands; and forecast discipline, because the fastest way to destroy the credibility that earns you allocation is to cry wolf — every inflated number you send a supplier to jump the queue this quarter is a withdrawal from the account you will need next year.

---

The through-line is a single reversal of perspective. In a constrained, fast-growing category, you are not a buyer selecting from eager suppliers; you are a customer competing to be chosen by suppliers who have more demand than they can serve. Once you accept that, the moves follow: earn allocation with credible, durable demand before you need it; separate the categories that decide your fate from the ones that don’t and match the strategy to the box; anchor for certainty and second-source for optionality, in that order; and govern the whole thing so it is still true a year from now. Do that and supply stops being the thing that caps your growth. Get it wrong and it will be the only thing anyone talks about — on the morning the line stops for want of a part you could have secured eighteen months earlier.

That is the work I do.

---

The process, in three stages

You do not choose a strategy for a category until you know where it sits. You do not execute a plan you have not chosen.

▼

▼

Each stage inherits the one above it. Nothing starts until the stage before has an answer.