Why I Do Not Begin with Just-in-Time
Supply chain suffers from a recurring temptation: to treat the whole discipline as a problem of cleanliness. Fewer buffers, smoother flows, tighter compliance, less visible waste. This habit of mind has produced useful practices, and it has also encouraged a narrowing of vision. Once the matter is framed too early as a quest for purity of flow, the commercial world is forced into a shape it does not naturally hold. Suppliers disappoint, customers change their minds, prices move, transport falters, promotions overshoot, and the future declines to stay tidy for our convenience.
I set out my own view in Introduction to Supply Chain, especially in Chapter 1 (“Primer”), Chapter 4 (“Economics”), Chapter 7 (“The Future”), and Chapter 8 (“Decisions”). The core idea is that supply chain is the practice of making economically sound commitments about the flow of goods while the future remains uncertain. Purchase orders, stock allocations, price changes, markdowns, routing choices, postponements, and capacity reservations belong to the same family of acts. They commit scarce resources today in the hope of a better outcome tomorrow. Once the subject is framed that way, inventory ceases to be the natural center of the theory. It becomes one lever among several.
Choice before doctrines
When choices come first, the proper unit of analysis is rarely the annual policy or the neat percentage on a dashboard. It is the next commitment. One more unit here, one later there, one more truck departure, one discount, one deferment, one transfer, one expediting decision. The governing question is economic and marginal: given everything already committed, where does the next slice of cash, capacity, time, or goodwill earn the best return? That question is wider than replenishment, and more demanding too, because it forces many levers to compete on the same ledger.
This also changes the role of uncertainty. On a factory line, variability is often a defect to be beaten down. In a commercial network, variability is the ordinary condition of life. Demand drifts, lead times stretch, suppliers fail, competitors change price, and external events intrude without asking permission. A company cannot purchase a deterministic future. It can only prepare for uncertainty by keeping worthwhile options alive and by valuing them correctly. Extra stock in the right place, spare capacity before a surge, a second supplier kept warm, or a decision delayed until better information arrives can all be sound economic choices. They may look untidy to someone who sees only the local process. They can be very sensible when seen from the balance sheet.
For the same reason, I have no wish to turn forecasts into little sovereigns. I use (probabilistic) forecasts heavily; I simply refuse to worship them. They can be useful inside the machinery. They become harmful when promoted into public commandments. Similarly, a service-level target cannot tell you what one more point is worth; a safety-stock formula that treats each SKU in isolation cannot arbitrate among competing uses of capital. Once these intermediates begin to rule the organization, the firm starts optimizing the laboratory glassware instead of the business itself.
JIT in its proper place
This is where just-in-time enters the discussion. Toyota describes JIT as making only what is needed, when it is needed, in the amount needed, within a synchronized production flow. Lean teaching adds pull, takt time, continuous flow, and production leveling. In the right habitat, especially inside manufacturing environments with disciplined processes, short lead times, and dependable inputs, this is a major achievement. It exposes dysfunctions that bloated inventories used to conceal. It shortens delays. It forces quality problems into the open. Any serious observer should grant that much immediately.
My disagreement begins when this manufacturing discipline is promoted into a general philosophy of supply chain. JIT naturally takes waste elimination and flow discipline as its compass. I take the firm’s risk-adjusted economic return on scarce resources as mine. The distinction matters. A flow can look beautifully lean in operational terms and still be economically brittle once the business depends on long replenishment lines, seasonal demand, promotions, political supply risk, or expensive stockouts. Lower inventory is often beneficial. It is not a self-justifying goal. There are many situations in which the rational choice is to hold more stock, buy more capacity, or pay to keep an alternative source available.
Inventory, in the broader view, is neither virtue nor vice. It is one form of optionality. The same is true of spare capacity, postponement, dual sourcing, and the ability to defer a binding decision until tomorrow’s information is better than today’s. The serious question is always the same: what is the value of keeping this option open, and what premium are we paying for it? JIT tends to treat slack as an admission of process failure. I am more willing to see slack as an asset when it buys resilience or allows the firm to exploit asymmetries in the future. A business that keeps goods generic for longer, delays final allocation, or maintains a modest buffer in one vulnerable node may look impure to a JIT purist and yet be much wiser economically.
The same divergence appears in the old language of push and pull. Too much supply-chain debate assumes that once one answers “pull” or “push”, the rest of the method follows. I do not believe this. A decision can be triggered by recent sales, by forecasts, by prices, by capacity constraints, by contractual obligations, or by some combination of all of them. The interesting question is whether the decision is economically sound under uncertainty. That requires probabilities, not a small theater of tidy scenarios, because the tails often matter more than the median and because the granularity of the decision should determine the granularity of the prediction.
The broader craft
Once one starts from decisions rather than departments, pricing enters the picture immediately. A markdown on a slow mover, a higher price on constrained capacity, a discount timed to free warehouse space, these are not foreign to supply chain. They change what moves, where, and when. Traditional JIT discourse tends to treat demand as something operations receives from the outside world and then serves as elegantly as possible. Real firms shape demand every day. Price is one of the strongest levers they possess, and a supply chain that ignores it leaves a decisive part of the problem outside the room.
Another consequence concerns the human role. Lean is right to respect operators and direct observation. I do too. No supply chain improves without close contact with the people who actually run docks, routes, replenishment calendars, and factory constraints. Yet modern supply chains generate far too many micro-decisions to leave daily arbitration inside spreadsheets, alert queues, and visual boards. The clerical work should disappear into software. People should spend their intelligence on choosing the economic model, testing the numerical recipe, correcting the semantics of data, and revising valuations when the world changes. The routine commitments themselves should be emitted automatically, logged, and later audited against outcomes.
Nothing in this requires hostility toward lean. I borrow from it whenever it earns its keep. Smaller batches can be profitable. Shorter lead times usually help. Better quality always helps. Build to order can be excellent in parts of a network. Direct observation remains indispensable. What I reject is the habit of turning such practices into articles of faith. They are heuristics. A good heuristic deserves to be encoded, measured, and retained only if it improves the business over time. If it does, keep it. If it does not, retire it without sentimentality.
For that reason I do not begin with just-in-time, and I certainly do not end there. Supply chain asks for a wider discipline: one that sees the firm as allocating scarce resources under uncertainty, one that values options instead of purity, one that treats forecasts and KPIs as instruments rather than masters, and one that accepts automation as the normal medium of modern decision-making. Under that view, a beautifully lean flow is welcome when it is justified. An unfashionable buffer is welcome too, if it earns its place. What deserves judgment, in the end, is the quality of the commitments written into the business and the money those commitments make or save over time.
