Every few years a company reorganises its procurement function, and the debate is always framed the same way: should buying be centralized under one team, or decentralized to the business units who actually feel the need? It's a tidy question. It's also the wrong one.
Because the honest answer to "centralized or decentralized" is: yes. Different categories, different speeds, different stakes — they don't all want the same model. The organisations winning at procurement stopped picking a side and started splitting the decision. The industry has a name for it now: center-led.
01 — The case for centralizingScale is the whole argument.
Centralized procurement concentrates buying authority in one team. The payoff is leverage: when all of an organisation's demand flows through a single point, you negotiate as one large buyer instead of forty small ones. You standardise specs, consolidate suppliers, enforce policy, and finally see your spend.
The numbers behind that are not small. Bringing fragmented spend under management typically unlocks 5–15% in savings[4]. Walking into a negotiation with real, consolidated spend data improves outcomes by 15–25%[6]. And standardising low-value buying through catalogs and automation cuts the cost of processing each transaction by 50–70%[7].
Fig. 1 — What scale unlocks
The dividend of buying as one
The cost of centralizing is just as real, though — it's slower, it can feel bureaucratic, and a central team rarely knows a local market the way the people standing in it do.
02 — The case for decentralizingSpeed, and the leak that comes with it.
Decentralized procurement pushes authority out to the business units. Decisions are fast, local knowledge is respected, and urgent or specialised needs get met without a queue. For a fast-growing company, that agility is often the right call early on.
But agility has a bill, and it arrives as fragmentation. When everyone buys independently, the same item gets purchased at five different prices from five different suppliers. Volume leverage evaporates. Supplier counts balloon. And compliance quietly erodes — somewhere between 20% and 40% of enterprise spend ends up happening off-contract[2], with The Hackett Group estimating organisations lose 5–16% of their negotiated savings to this maverick buying[3].
Most of that damage hides in the tail.
Fig. 2 — The 80/20 inversion
A fifth of the money. Four-fifths of the suppliers.
Centralization captures leverage and loses speed. Decentralization captures speed and loses leverage. Neither pays for the other.
03 — The tradeoff, mappedNo model wins every dimension.
Put the two pure models against the things procurement is actually judged on — cost leverage, speed, compliance, local fit, spend visibility, agility — and the shape of the problem becomes obvious. Each pure model is brilliant on its own axis and weak on the other's. The center-led model doesn't top any single axis; it refuses to collapse on any of them.
Fig. 3 — Three models, six dimensions
Center-led is the model that doesn't have a weak side
Center-led: centralize the strategy, decentralize the buy.
Center-led procurement resolves the tension by splitting the decision instead of the org chart. A central team — often a Center of Excellence — owns everything strategic: category strategy, framework contracts, supplier vetting, governance, analytics, and the buying platform itself. The business units keep what they're best at: choosing within approved options, moving fast, handling the genuinely local and urgent. KPMG's research shows this is where most modern procurement organisations have landed[1].
- Category strategy & sourcing playbooks
- Framework contracts & negotiated pricing
- Supplier vetting, risk & onboarding
- Spend analytics, governance & compliance
- The buying platform, catalogs & guardrails
- Day-to-day requisitions & approvals
- Choosing within pre-approved catalogs
- Urgent & genuinely local supplier needs
- Receiving & delivery sign-off
- Flagging gaps back to the COE
Done well, the business unit barely notices the "central" half exists. They open a catalog, pick what they need, and check out — fast, the way decentralization promised. They just happen to be buying at negotiated prices, from vetted suppliers, with full visibility, the way centralization promised. The split is invisible at the point of purchase, which is exactly the point.
05 — Why this is finally practicalTechnology is what makes the split real.
Center-led isn't a new idea — it's an old one that used to be impossible to run. For decades, "central sets the contracts, local does the buying" fell apart in practice, because there was no way to put the negotiated contract in front of the person making the purchase at the moment they made it. So they called a supplier they knew, and the leverage leaked.
That's the constraint technology removes. Catalogs, punchout into ERP, and guided buying place the approved, contracted option directly inside the requisition screen. The compliant path becomes the easy path — and when the compliant path is also the fastest one, maverick spend stops being a discipline problem and starts being a non-event.
If buying through the system is slower than going around it, people go around it. Center-led only works when the centre makes local buying faster, not harder.
This is the through-line: the centralize-or-decentralize debate was always really a debate about a missing tool. Once the negotiated contract can travel to the point of purchase, you no longer have to choose between control and speed. You hold the contract centrally and let everyone buy locally — at the same time.
06 — The real questionNot who holds the power. How you split the decision.
So retire the old question. "Centralized or decentralized" forces a trade you don't have to make. The better question is sharper and harder: which decisions belong at the centre, and which belong at the edge?
Strategy, contracts, and standards belong at the centre, where scale lives. Execution belongs at the edge, where speed lives. Get that line right, give the edge a system that makes the right choice the easy one, and the century-old tug-of-war simply stops being a tug-of-war.
The best procurement organisations aren't centralized. They aren't decentralized either. They're deliberate about which is which.