An opinionated introduction to supply chain
When you see a title like Introduction to Supply Chain, you expect a survey of prevailing ideas, a neutral primer. Mine is not that. It had to be opinionated. After nearly two decades spent operating supply chains with Lokad—across retail, aerospace, manufacturing, and more—I no longer believe a “vanilla” introduction helps practitioners. The book is available to read in full, but the spirit of it can be stated plainly: start from reality, not rituals; judge choices in coins, not in proxies.
Profit is the yardstick
Supply chain is the discipline of allocating scarce resources under uncertainty. In practice, that means we exist to raise a firm’s risk‑adjusted return on capital. Everything else—service level, inventory turns, lead times, even sustainability metrics—matters insofar as it improves that return over time. The alternative is to optimize for elegant dashboards that leave money on the table. This stance is not contrarian for its own sake; it is the only stance that consistently pays its way when decisions are priced and audited.
From plan‑first to decision‑first
The mainstream playbook begins with consensus plans, targets, and a “single number” forecast. But the world does not hold still for our numbers. What we can hold constant is the discipline of making better decisions today while preserving room to maneuver tomorrow. Planning is useful only to the extent it sharpens concrete commitments—what to buy, make, move, and price—and only when those commitments are continually revised as information arrives. Treating S&OP as a monthly ceremony that blesses a forecast is an expensive distraction from this decision work.
Automation must carry the pen
Software earns its keep when it commits—when it places orders, sets allocations, moves prices—unattended, with audit trails and the humility to halt itself when a rule is breached. “Decision support” that floods human teams with suggestions and alerts is just a modern UI around yesterday’s clerical workload. The point is not to remove judgment, but to move it upstream: human judgment defines the economics and constraints; the machine applies them, every night, at full scope.
The comfortable classics are usually costly
Formulas and heuristics that look harmless in isolation—safety stock tables, ABC splits, service‑level targets—tend to ignore portfolio effects, fat tails, and the hard link between working capital and opportunity cost. They institutionalize arbitrary parameters and train teams to chase proxy KPIs. In practice, these “classics” are often the most expensive habits a company keeps, precisely because they feel so reasonable. The evidence against them has been in front of us for years.
Forecasting is a servant, not a shrine
I like good forecasts as much as anyone. I just don’t worship them. Better forecast accuracy can worsen decisions if it nudges an organization to over‑commit or to optimize to a KPI that has drifted away from profit. Forecasts should be probabilistic, tied to the policies they inform, and judged by the decisions they enable—not by a score that flatters a chart but starves the P&L. If a worse‑looking forecast helps a better decision, it is the better forecast.
Commercial levers are inside the perimeter
Pricing, promotions, and assortment shape demand and the flow of goods; treating them as “someone else’s problem” makes the rest of supply chain a game of catch‑up. In practice, procurement, replenishment, allocation, and pricing must be optimized together. Most teams say this; few actually run one recipe that binds these levers into coherent commitments. They should.
Keep options open—waiting is a decision
Not every choice should be rushed. Often the most profitable move is to wait, buy a little information, and act when the expected return clears a sensible bar. Organizations under‑invest in this kind of optionality because it is cognitively hard and politically unfashionable; it looks like indecision. Software can fix both problems by making the option value explicit and by standardizing when we act and when we hold.
Ledgers should not be the brain
Enterprise software comes in three useful kinds: the systems that record transactions, the systems that report on them, and the systems that decide. Confusing these roles is how companies end up asking their ledger to plan, and their dashboard to optimize. Keep the record‑keeping simple and reliable. Let reporting do what reporting does. Put a separate decision engine in charge of routine commitments, where economics and uncertainty belong.
Beware the knowledge theater
Our field is awash with white papers, case studies, quadrants, and market maps that supply the appearance of certainty while avoiding the only test that matters: uplift in coins, measured against a live baseline. Incentives being what they are, many artifacts are infomercials with footnotes. If a method cannot survive a head‑to‑head, full‑scope run against the incumbent—even for a few weeks—it is not ready for your operation.
Progress looks like experiments, not ceremonies
Real change feels less like a big‑bang cutover and more like repeated, disciplined experiments. Run the new recipe in shadow; let it compete with the incumbent on the same data and the same days; keep what wins. This is how you de‑risk novel ideas while giving them a fair chance to prove themselves. It is also how you learn faster than competitors who rely on meetings to substitute for evidence.
The human role, upgraded
If software carries the pen, what do people do? They become stewards of meaning and money: curating the frames the machine uses to understand the business; publishing the economics (rebates, penalties, capital charges) that shape trade‑offs; maintaining constraints that reflect physical reality and policy. This is far richer work than line‑editing suggestions. It is how supply‑chain teams move from clerical equilibrium to strategic relevance.
Why call this an “introduction” at all?
Because newcomers deserve a clear starting stance. A neutral survey would pretend disputes don’t exist—and strand you with methods that were convenient before modern computing, but are poor guides now. The mainstream keeps its ceremonies; trucks still move, so nothing looks broken. Yet the opportunity cost is enormous. If you are willing to align economics, computation, and organization, the gap between what you have and what you could earn is wide open.
This is the introduction I wish I had found when I started my supply chain journey: not a museum of methods, but a map of practice that holds up under pressure. If it sounds opinionated, it is. It is also what has worked—repeatedly, at scale—when real money, real goods, and real customers were on the line.
