The fastest way to lose money you never planned to lose is to treat buying as paperwork. A late component halts a production line; a single supplier going dark strands an entire product; a clever volume discount quietly locks you into stock you can't shift. Supply chain, procurement and sourcing are the three overlapping disciplines that keep that from happening, and the leaders who run them well rarely get credit, because their work shows up as the absence of a crisis.

The quick version

  • Sourcing is choosing who you buy from, finding, evaluating and selecting suppliers. Procurement is the wider process of acquiring what you need: sourcing plus negotiating, contracting, ordering, paying and managing the relationship. Supply chain is the whole flow, suppliers, making, moving and delivering, that gets a product from raw material to customer.
  • Not every purchase deserves the same effort. Peter Kraljic's portfolio model sorts what you buy by profit impact and supply risk, so you spend your scarce negotiating energy where it actually matters.
  • Cheap to buy is not the same as cheap to own. Total cost of ownership counts delivery, training, downtime, maintenance and disposal, the costs hidden below the price.
  • Small wobbles in customer demand get amplified into wild swings further up the chain, the bullwhip effect. Better information, not bigger buffers, is the fix.

The idea in depth: three words people use interchangeably

The terms blur together in everyday speech, and the blur causes real confusion in meetings. Keep them straight: sourcing answers "who do we buy from?"; procurement is the end-to-end process of acquiring goods and services, it contains sourcing but also covers negotiation, contracts, purchase orders, payment and supplier management; and supply chain is the entire network and flow that turns inputs into a delivered product. Sourcing sits inside procurement; procurement sits inside the supply chain.

Which is why it pays to be precise about which one you're actually fixing. "We have a supply-chain problem" might mean a supplier was chosen badly (sourcing), that contracts give you no leverage (procurement), or that goods can't move fast enough (logistics, deeper in the chain). Each has a different owner and a different cure. Name the layer and you've already done half the work of fixing it.

flowchart LR
  A(["Sourcing
who do we buy from?"]) --> B(["Procurement
acquire: negotiate, contract, order, pay"])
  B --> C(["Supply chain
supply → make → move → deliver"])
  C --> D(["Customer"])
Sourcing nests inside procurement, which nests inside the wider supply chain. Leaders Loop

Treat different purchases differently: the Kraljic matrix

The single most useful idea in procurement is that you should not buy a year's supply of pencils the way you buy the one bespoke part only one factory on earth can make. That insight was made rigorous by Peter Kraljic, then a McKinsey director, in "Purchasing Must Become Supply Management" (Harvard Business Review, September 1983), the paper that argued procurement should stop being a clerical function and start being a strategic one. His portfolio model sorts everything you buy on two axes: profit impact (how much this spend matters to your costs and results) and supply risk (how hard it would be to get if your supplier vanished).

Those two axes make four quadrants, and each calls for a different strategy. Non-critical / routine items (low impact, low risk, office supplies) you simplify and automate; don't waste a buyer's week on them. Leverage items (high impact, low risk, many willing suppliers) are where hard negotiation and competitive tendering pay off, because you hold the power. Bottleneck items (low impact, high risk, a niche part with one source) call for securing supply and finding alternatives, even at a price premium. And strategic items (high impact, high risk) need deep, partnership-style relationships and joint planning, because you and the supplier are mutually dependent.

flowchart TD
  K(["Sort each purchase by
profit impact & supply risk"]) --> R(["Routine
low impact, low risk
→ simplify, automate"])
  K --> L(["Leverage
high impact, low risk
→ tender, negotiate hard"])
  K --> B(["Bottleneck
low impact, high risk
→ secure supply, find alternatives"])
  K --> S(["Strategic
high impact, high risk
→ partner, plan jointly"])
Kraljic (1983): sort spend by profit impact and supply risk, then match the tactic to the quadrant. Leaders Loop

In practice, take your top spend categories and drop each one into a quadrant before your next supplier review. Where it lands tells you where to point effort: tender the leverage items, derisk the bottlenecks, nurture the strategic few, and stop hand-crafting the routine.

An honest limitation. The matrix is a lens, not a verdict. Critics have long noted that "profit impact" and "supply risk" are judgement calls, that the model says little about the supplier's view of you (a small buyer of a strategic item may have no leverage at all), and that a category can sit on a boundary or move quadrant when markets shift. Use it to structure the conversation, not to end it, and revisit the placements as conditions change.

Why orders swing wildly upstream: the bullwhip effect

Here is the phenomenon that surprises every new operations leader. End customers buy at a fairly steady rate, yet orders placed by the retailer, then the distributor, then the factory, then the factory's suppliers grow more and more erratic the further upstream you go. A gentle ripple at the till becomes a tidal swing at the raw-material supplier. This is the bullwhip effect, named and dissected by Hau Lee, V. Padmanabhan and Seungjin Whang in "The Bullwhip Effect in Supply Chains" (Sloan Management Review, 1997). Their headline example: Procter & Gamble found that while parents bought Pampers at a steady clip, distributor orders varied more, and P&G's own orders to its suppliers (such as 3M) swung more wildly still.

Their key finding is that this is not stupidity, it is rational behaviour by each link reacting only to the orders immediately downstream. They trace it to four causes: demand-forecast updating (everyone re-forecasts off their customer's orders, not real demand); order batching (ordering in big periodic lumps rather than little and often); price fluctuation (promotions trigger forward-buying that distorts the signal); and rationing and shortage gaming (when supply is tight, buyers inflate orders, then cancel). Each is a sensible local decision that produces a destructive global pattern.

The bullwhip isn't caused by bad people making dumb calls, it's caused by good people reacting to the only signal they can see.

The cure, then, is to attack the information problem rather than the inventory problem. Share real point-of-sale demand up the chain so everyone forecasts off the same truth; order smaller and more often to cut batching; replace promotional spikes with steadier "everyday" pricing where you can; and allocate scarce stock by past sales rather than current orders, so gaming stops paying. The classic teaching tool here is MIT's Beer Distribution Game, where players who can't see real demand reliably generate the bullwhip themselves, a humbling demonstration that the structure, not the person, creates the chaos.

A worked example

Take a mid-sized outdoor-gear brand, call it Ridgeline, that assembles backpacks. (Illustrative figures throughout; this is a teaching example, not real accounts.) Two purchases dominate its spend: a custom waterproof zip made by a single specialist mill in one country, and standard nylon fabric available from dozens of suppliers worldwide.

Run them through Kraljic. The nylon is a leverage item: high spend, many sources, Ridgeline holds the power, so the buyer runs a competitive tender and lands, say, an illustrative 8% unit-cost cut simply by making three suppliers compete. The custom zip is a bottleneck (edging toward strategic): modest spend but no alternative source, so a price-only fight is beside the point. The right move is to secure supply, a longer contract, a second source being qualified in parallel, a small buffer of zips, even though none of that lowers the sticker price. Treating both the same way would mean either squeezing the irreplaceable zip supplier until they drop you, or lazily single-sourcing the commodity nylon and leaving money on the table.

flowchart TD
  A(["Map Ridgeline's spend on Kraljic"]) --> B{"Which quadrant?"}
  B -->|"Nylon: high spend, many sources"| C(["Leverage → tender 3 suppliers
~8% unit cost saved (illustrative)"])
  B -->|"Custom zip: low spend, single source"| D(["Bottleneck → secure supply,
qualify a 2nd source, hold a buffer"])
  C --> E(["Lower landed cost"])
  D --> F(["Lower risk of a line-stopping stockout"])
Same company, two purchases, two opposite strategies, because impact and risk differ. Leaders Loop

Then a demand wobble tests the chain. A magazine feature triples Ridgeline's retail sales for one month. If each link reacts only to the order in front of it, the bullwhip kicks in: the warehouse over-orders, the assembler over-orders zips, the mill scrambles, and three months later Ridgeline is sitting on a mountain of zips for a spike that has long passed. The fix isn't a bigger zip buffer; it's sharing the actual retail-sales data with the mill so it can see the spike for what it was, temporary, and not amplify it. Information up the chain beats inventory at the end of it.

Frequently asked questions

What's the difference between procurement and sourcing?

Sourcing is one part of procurement. Sourcing is specifically about identifying, evaluating and selecting suppliers, deciding who you buy from. Procurement is the whole acquisition process around that decision: defining the need, sourcing, negotiating, contracting, raising purchase orders, receiving, paying, and managing the supplier afterwards. In short, all sourcing is procurement, but not all procurement is sourcing.

Is the cheapest supplier always the right choice?

No, and conflating price with cost is the most common procurement mistake. Total cost of ownership (a core CIPS concept) counts everything the purchase costs over its life: delivery, installation, training, energy, maintenance, downtime, support and disposal, the hidden part of the iceberg beneath the purchase price. A cheaper item that breaks often, or that nobody can support, frequently costs more to own than a pricier, reliable one.

How do I make my supply chain more resilient?

Start by knowing where you're exposed. Single-source bottleneck items are the obvious risk, qualifying a second supplier, holding a sensible buffer, or designing the part so an alternative could substitute all reduce it. Beyond that, resilience comes from visibility: knowing who your suppliers' suppliers are, and sharing real demand data so shocks aren't amplified by the bullwhip. Resilience usually costs a little margin in calm times; the question is how much insurance the risk justifies.

What does "strategic sourcing" actually mean?

It's sourcing treated as an ongoing, analytical discipline rather than a one-off purchase. Instead of buying reactively when something runs out, you analyse a whole spend category, understand the market and your own demand, segment what you buy (Kraljic is the usual starting frame), and build a deliberate supplier strategy for each segment. The "strategic" part is the shift from transaction to plan, exactly the move Kraljic argued for in 1983.

Why do my orders to suppliers swing more than my actual sales?

That's the bullwhip effect, and it's structural rather than personal. If you re-forecast off your incoming orders, batch your ordering, react to promotions, or inflate orders when supply looks tight, you amplify the signal, and so does everyone upstream of you. The cure is better shared information (real end-demand data), smaller and more frequent orders, steadier pricing, and allocating scarce stock by historical sales rather than current orders.

Related in the Toolkit

How you plan demand and supply together (S&OP & demand planning) is the discipline that most directly tames the bullwhip, and the way you map and re-engineer the flow of goods (process design & mapping) is how you find the waste a stressed supply chain hides.

Where to go next