What is a procurement operating model?

An operating model is usually presented as a structure diagram. The structure is the least interesting part of it and it is downstream of six decisions. Where those decisions have not been made explicitly, the organisation still has an operating model. It just has an accidental one, assembled from whatever people did last time.

1. 01MandateWhat spend procurement is accountable for, what it is not, and what authority it has over the spend it covers. The failure here is a mandate that says procurement is involved in everything and can decide nothing. No amount of process design fixes that, because it is not a process problem.
2. 02Decision rightsThresholds, delegated authority, who can commit the organisation at what value, what route a sourcing decision takes and who can stop it. This is the component that determines whether the model is real, and it is the one most often left in a policy document nobody has opened since it was issued. In a public body it lives in the scheme of delegation or the financial regulations, which means changing it needs a committee and a lead time.
3. 03The category boundaryWhich categories are bought once for the whole organisation, which are bought locally against a central agreement, and which are genuinely left alone. Not a principle. A list, with a named owner against each line.
4. 04Roles, and the time attached to themCategory ownership, sourcing, contract and supplier management, operational buying, systems and data. Roles matter more than boxes. A category manager who spends four days a week processing requisitions is not a category manager, whatever the chart says, and the model has to say what protects the other three days.
5. 05Process and technologyHow something gets bought from need to payment, which parts are standard, which are allowed to vary, and what carries it. Everything the model permits to vary eventually does.
6. 06InterfacesWhere procurement meets finance, legal, IT and the operational business, and at what point in a decision. Most of the delay people attribute to procurement is sitting in a handover, and handovers belong to nobody unless the model gives them to someone.

Answer those six and the plausible structures narrow to two or three. Draw the structure first and the six get answered by default, in the first month after the reorganisation, by whoever is under the most pressure that week. What the model is then measured on is a separate question, and one worth settling at the same time.

The centralisation question absorbs a disproportionate share of the debate, usually as a question of principle. It is a real question of principle where budgets are devolved, because it is a question about where power sits, and that gets settled above procurement. But it should not be settled only that way. Category by category, four tests do most of the work.

Leverage

Does buying this together materially change the price or the terms? Concentrated supply market, common requirement, yes. A market where local suppliers compete on availability and response rather than price, often not.

Requirement variation

Can one specification serve the whole organisation without someone quietly buying around it? Where the requirement genuinely differs by site, service or clinical need, a central contract produces workarounds rather than compliance, and the workarounds are invisible until someone counts them.

Proximity to the decision

How much operational knowledge does a good buying decision need, and how quickly? Some categories reward market expertise. Others reward being in the room when the requirement is written.

Who holds the budget

The test people leave out, and the one that decides whether the other three matter. You cannot run a category centrally when the money sits with a divisional managing director or a directorate finance lead who has not agreed to it. Either the budget moves, or the mandate is negotiated with the person holding it, or the category stays local and the model says so honestly.

Run the tests and most organisations land on a hybrid, because most organisations have categories that answer differently. The output is not a label. It is a list, category by category, with a named owner and a stated route, and it is the single most useful artefact a redesign produces.

The centre-led model that works is the one that has been specific about which categories it leads.

Four shapes cover most of what organisations actually do. Each has a characteristic way of breaking, and knowing the failure mode is more useful than choosing the shape.

1. 01CentralisedOne team buys everything above a threshold. Strongest leverage and the clearest accountability. Fails when it becomes a bottleneck: cycle time climbs, the business finds routes around it, and the compliance figure that looked healthy turns out to be measuring only the spend that came through the door.
2. 02Centre-led with lead buyersA central team sets the category position; people in the business execute against it. Works where the category boundary is explicit. Fails when the lead buyer role is a dotted line onto someone appraised and promoted locally, because the local objectives win every time.
3. 03Centre of excellence with devolved executionA small central team holds capability, standards and the difficult negotiations. Cheap and credible on paper. Fails through lack of throughput: with no delivery of its own the centre becomes advisory, and an advisory function is one people consult when they already agree with it.
4. 04Procurement in a shared serviceTransactional buying and often accounts payable sit in a shared service or an outsourced tower, with a commercial team retained. Genuinely effective at the transactional end. Fails at the seam: the retained team loses visibility of what is actually being bought, and the service is measured on throughput rather than on whether the buying was any good.

The reporting line is a separate decision and it is not neutral. Reporting to the finance director buys credibility for the savings numbers and makes it harder to be in the room when a requirement is being written. Reporting to the chief operating officer gets procurement into requirement setting and makes its numbers more contested. Neither is wrong. What is wrong is not noticing that the choice was made.

The failures are consistent enough to be predicted, and almost none of them are structural.

1. 01Decision rights left where they wereThe structure is redrawn and the thresholds are not. Nothing changes, because the thing that governed behaviour was never the chart.
2. 02No interim modelThe design describes the destination and nothing describes the period in between. The function operates on the old model while being measured against the new one, which is a good way to lose the argument for the new one.
3. 03Category ownership without category timeOwnership assigned on top of an existing operational workload. The operational work has a deadline and the category work has a date. The deadline wins every week.
4. 04People treated as fungibleThe design assumes roles can be filled. In practice grading constrains what a category manager can be paid and therefore who applies, any structural change triggers a consultation with its own timetable, and the uncertainty window is exactly when the two people you most wanted to keep take calls from recruiters.
5. 05The mandate never actually agreedSigned off by an executive team, never accepted by the budget holders it applies to. It surfaces the first time a significant purchase goes around the model and nobody senior enforces it.
6. 06Nobody owns the model itselfDesigned, launched, then not maintained. A restructure later it describes an organisation that no longer exists, and everyone has stopped consulting it.

A redesign that opens with six months of workshops produces a document. A shorter sequence starts with evidence and settles the arguments that would otherwise be settled by seniority.

1. 01Establish what is actually happeningSpend by category and supplier, contracts and their expiry, and who is currently making buying decisions at what value. With one ledger and a usable accounts payable extract this is weeks. With several entities, no supplier parent hierarchy and a contract register last updated two years ago, the spend is weeks and the contract picture is considerably longer, because it is a document hunt rather than a query.
2. 02Fix the mandate with the people it constrainsNot with the executive team alone. The conversation that matters is with the budget holders, and it is a negotiation: procurement is asking for authority and has to say what it is offering in return, usually speed, market knowledge or taking work off them.
3. 03Decide the category boundaryApply the four tests and produce the list. Almost everything downstream depends on this and almost nothing depends on the org chart.
4. 04Write the decision rightsDelegated authority, approval route, escalation path. Make them shorter than the ones being replaced, and find out early what the audit committee and the external auditor will accept, because in a public body that is the binding constraint rather than the appetite of the executive.
5. 05Size the team to the workOnly now is structure a useful conversation, because you know what the function is doing and for whom.
6. 06Write the interim model downHow the function operates from next month until the target is real, who holds what in the meantime, and what it is fair to measure it on while the change is happening.