Designing the purchasing function in a manufacturer: who buys, who chases and who decides

The buyer negotiating the contract cannot also chase the line

The single most useful split is between the person responsible for the commercial position on a category and the person responsible for material arriving on Thursday. Combined in one role, the urgent always displaces the important, and the strategic work never happens because a stopped line will always win the morning. Separating them changes the job descriptions, the measures and the type of person recruited. It also creates a handover risk, so define clearly which role owns forecast communication, which owns the delivery escalation, and at what point a recurring operational problem becomes a commercial one.

Where supplier quality engineering belongs

Supplier quality sits awkwardly and the placement matters. Inside quality, it retains independence and the authority to reject, but risks being seen as an inspection service rather than a development function. Inside procurement, it gets close to sourcing decisions and supplier development, but a reporting line into the function that owns cost creates obvious pressure when a rejection is inconvenient. Either can work provided the escalation route bypasses the commercial owner and the reject decision cannot be overturned informally. What does not work is leaving it unassigned and expecting buyers to do it.

Centralising a category without stranding the plant

Grouping spend across sites gives negotiating weight, consistent terms and one owner of a supplier relationship. It also removes the local knowledge that made the previous arrangement work: which supplier will take a call at night, which one knows the process, which one has the fixture. Centralisation fails when it is imposed on categories where responsiveness matters more than price, and when plants lose the ability to solve their own problems. The usual settlement is central ownership of the agreement and local ownership of the call-off, with the plant able to escalate quickly when the central choice is not delivering.

The engineering interface decides most of the cost

Material cost is largely fixed by the time a drawing is released: the material chosen, the tolerances specified, the finish called out and the components selected determine which suppliers can bid and what the part will cost. A purchasing function positioned downstream of design can negotiate the margin and nothing else. Placing sourcing engineers inside development teams, giving purchasing a voice at design review, and making component availability a design criterion moves influence to where it matters. This is an organisational choice about reporting and timing, not a matter of asking engineers to be more cooperative.

Sizing the function against what it actually buys

Headcount driven by transaction volume produces a team busy processing orders and incapable of managing suppliers. Better drivers are the number of suppliers requiring active management, the number of new parts introduced in a period, the proportion of spend under agreement, and how technically demanding the purchased content is. Transactional work can often be consolidated into a shared service or automated once agreements exist, which frees capacity rather than saving headcount. Whatever structure emerges, name a single owner for every significant supplier, because unowned relationships fail quietly. An organisation chart with unattributed suppliers on it is describing an intention rather than a structure.

Frequently asked questions

Should a multi-plant manufacturer centralise purchasing?

Partially, and by category rather than wholesale. Categories with common specifications across sites, significant spend and stable requirements benefit from a single agreement and a single relationship owner. Categories where the specification is site-specific, the volumes are small or responsiveness is the main requirement are usually better left local. The hybrid that works has central ownership of agreements and supplier relationships, local ownership of scheduling and day-to-day resolution, and an explicit escalation path when the central arrangement is failing a plant.

What skills should we recruit for as the function matures?

Early on, reliability and follow-through matter most, because material arriving is the job. As agreements come into place, the constraint shifts to commercial analysis, contract literacy, cost modelling and the technical confidence to challenge a specification. Technically trained buyers who can read a drawing and hold their own with engineering are disproportionately valuable in a manufacturing setting. The hardest gap to fill is usually someone who can run a category strategically while remaining credible on the shop floor.

How do we stop plants buying around a central agreement?

Find out why they are doing it first, because the usual reason is that the central arrangement is not meeting a real need — availability, responsiveness, a technical requirement nobody captured. Enforcement without fixing that just moves the workaround somewhere less visible. Make the compliant route genuinely easier than the alternative, give plants a fast escalation when it fails, and reserve system controls for the cases where the underlying need has been addressed and the behaviour continues anyway.