The price on the invoice is a negotiating outcome. The cost behind it is the economics — and only one of the two tells you what to do next.
Most enterprises can state to the cent what their largest supplier charges. Very few can state, within a defensible band, what it costs that supplier to serve them. The distance between those figures is somebody's margin, and the interesting questions are whose, what it buys, and whether it will still be there in three years.
Procurement is usually organised around the first number: annual negotiation, price variance against budget, a savings target as a percentage. That measures negotiating outcomes, not why the price is what it is — the only information that helps a leadership team decide whether to negotiate harder, redesign the specification, qualify a second source, or leave the arrangement alone.
The discipline is closer to cartography than haggling: map the chain from raw input to the moment your customer pays, identifying where value is captured and what it buys at each step. Michael Porter's value chain, extended to the wider value system a firm operates in, remains the established frame.
The Strategic Context
For most enterprises the cost of goods is largely bought, not made. Design, assembly, integration and service sit inside; most of the unit cost sits with suppliers and their suppliers. The enterprise therefore has least influence, and least information, over the largest component of its own economics.
Two pressures make that gap expensive. Input volatility — currency, freight, energy, regulated charges — transmits with lags, and increases pass through faster than decreases. And consolidation upstream converts a competitive supply market into a concentrated one quietly, two tiers from anything a buyer watches.
There is also a visibility asymmetry. A significant supplier understands your cost structure better than you understand theirs: they see your volumes, specifications, engineering changes and urgency. You see an invoice.
What a Supplier's Margin Is Assumed to Be
The most common misreading is that margin held elsewhere in the chain is money the enterprise loses. Margin buys something: risk absorption, inventory carried on someone else's balance sheet, aggregation across many small buyers, warranty exposure, capital sunk in tooling, a scarce capability. Remove an intermediary and you inherit the function along with the margin — usually without their scale, always with their working capital.
The second is that cost transparency means open book. An open-book arrangement gives you a supplier's allocation of their own costs — an accounting artefact reflecting their choices about overhead absorption. Useful, but not an independent estimate, and a buyer who has never built one cannot challenge it.
The third is that a cheaper input produces a lower cost. It does where the specification tolerates it. Where it does not, the difference reappears as scrap, rework, warranty and recovery effort — in a different budget from the one that recorded the saving [Related article: What Is Your Quality Failure Costing You — and Can Finance Produce the Number?].
Reframing the Issue
Replace "what will they charge us next year" with "where along this chain is value captured, by whom, and what protects it". Two categories follow, with different responses.
Defensible margin rests on something real: a scarce input, a regulated licence, technical capability, capital intensity, switching costs, or risk the supplier actually carries. Attacking it is a multi-year capability programme, an acquisition, or a mistake.
Informational margin exists because the buyer has not looked. It rests on asymmetry alone and is recovered with analysis rather than capital — the cheapest margin in the enterprise to go after, and the one most reliably left untouched because nobody owns the question.
Reading the Margin Stack
The method is decomposition. Break the delivered price into components you can estimate independently: the input's indexable commodity content and yield; conversion, in labour and machine hours and scrap; logistics; the cost of financing inventory between conversion and sale; the residual.
The residual is the useful line: what you are paying for everything you cannot name. A large residual is not evidence of exploitation — it may be where the technical capability sits — but it is evidence of a question the enterprise has not asked. A component resolving into two-fifths indexable material, modest conversion, freight at published rates and a residual of a third tells you nothing about fairness and everything about where to direct the next conversation.
Much of the data is observable without the supplier's cooperation: commodity indices, published freight rates, tariff schedules, typical yields, tooling amortisation, lead times. The point is not precision but to make the next discussion a comparison of estimates rather than a request for a discount.
The chain also has a financing dimension. Who funds the inventory, who holds obsolescence risk, and who is paid before whom determine how it behaves under stress — an enterprise financed by its suppliers or customers has a different exposure when volume falls than its margins suggest [Related article: Can Your Customers Fund the Business — and What Breaks When Volume Falls?].
The Analysis That Is Called Cost-Benefit and Usually Is Not
Sourcing, insourcing and vertical integration decisions are typically supported by a familiar method: list costs, list benefits, assign values, total the columns, proceed if benefits exceed costs. It has the shape of analysis and four defects.
No discounting. Costs are front-loaded and benefits spread across years, so comparing undiscounted totals favours whichever side arrives later — in an insourcing case, the benefit side. A dollar of benefit in year five is not a dollar today, and the cost of capital is the number that says so.
No stated horizon. Whoever selects the horizon selects the answer. Where it is never stated, it has been chosen implicitly by whoever built the model.
No sensitivity analysis. A single point estimate is presented as a result. The decision-useful output is which two or three assumptions the conclusion depends on, and how far each can move before it reverses.
A cost taxonomy that double-counts by construction. Direct, indirect, opportunity, tangible and intangible are not five categories but descriptors along two dimensions — attribution and measurability. A piece of equipment is both direct and tangible; total all five columns and it is counted twice. Opportunity cost is worse: not a line item but what the comparison is, the value of the alternative forgone, which belongs in the baseline. Add it to a column as well and it is counted twice again.
A defensible appraisal states the baseline and horizon, uses incremental cash flows discounted at the cost of capital, names its two or three critical sensitivities, and names the executive who owns the benefit. Non-financial factors are carried as explicit judgements rather than invented dollar values: a fabricated number is not more rigorous than a stated judgement, because it conceals where it came from. The same caution applies to reading cash movements as evidence of economics [Related article: Cash Arrives Fast. Is That Profitability, or Only That You Have Not Found Out?].
Decision Framework
| What the margin rests on | What it actually buys | The right move |
|---|---|---|
| Scarce or controlled input | Access | Secure supply; qualify substitutes; index the contract |
| Regulated licence or accreditation | Permission to operate | Pay it; do not attempt to replicate |
| Capital intensity or tooling | Someone else's balance sheet | Compare against your cost of capital |
| Aggregation and logistics | Scale you do not have | Buy it until your volume exceeds theirs |
| Risk and inventory carry | Variability you would otherwise hold | Price the risk before repatriating |
| Information asymmetry | Nothing | Build the cost model; renegotiate on evidence |
Three tests sit alongside it. The residual test: can we name what the unexplained portion of this price buys? The function test: if this supplier disappeared, who would do the work, at what fully loaded cost including working capital? The reversibility test: if we insource and are wrong, what does it cost to go back, and how long?
From Strategy to Execution
The immediate work is narrow and quick: take the three largest bought categories and build a rough cost model for each — days of effort, not months. Then bring the model to the supplier instead of a discount request, and ask about drivers, not price.
Over the medium term two capabilities need funding: a cost-engineering function that can decompose a price and hold its own technically, and a standard appraisal template with a mandatory baseline, horizon, discount rate and sensitivity section, so no significant sourcing decision reaches approval as a comparison of undiscounted columns.
The long-term question is structural: which steps in the chain the enterprise intends to own, which to partner on, and which to deliberately rent — decided on where value is captured rather than where the organisation chart stops.
Signals to Monitor
Price rises attributed to inputs whose published indices have not moved. Suppliers willing to take volume growth but unwilling to discuss cost drivers. Consolidation among your suppliers' suppliers, two tiers from where anyone looks. A residual line that widens year on year. Savings reported against last year's price rather than a cost model — negotiation, not economics.
Questions for the Leadership Team
- For our three largest bought categories, can anyone describe the supplier's cost structure without quoting the supplier?
- Which margins in our chain are defensible, and which exist only because we have not looked?
- When we last approved a make-or-buy decision, what horizon, discount rate and sensitivities did the paper state?
- Where in our appraisals is opportunity cost added to a cost column and also embedded in the baseline?
Closing Perspective
An enterprise can treat its supply chain as a series of prices renegotiated annually, or as a structure of value capture it can read. The first produces savings that are real, small and self-limiting, stopping at whatever the counterparty judges tolerable. The second produces decisions about where the business should sit in the chain.
The distinction shows up under pressure. When an input moves sharply, an organisation that holds a cost model knows within a week which price increases are justified and which are opportunistic. An organisation that holds only invoices finds out by paying, and calls the result a market condition.
About the author
Kevin Jogin is Founder & Principal Advisor at EraNorth. Meet the Founder.