Make-to-stock or make-to-order: where you park the risk

Comparison criteria

Criteria are stated explicitly and neither option is declared a winner: which one fits depends on the constraint that binds hardest in your operation.

The decoupling point is the actual decision

Framing this as two models hides the useful question, which is how far down your process a customer order reaches before it starts driving activity. Push that point to the shipping bay and you are holding finished goods. Push it back to goods-in and nothing moves until an order arrives. Between those extremes sit the arrangements most plants actually run: painted and stocked, assembled to order; blanks held, machined to order; bulk made ahead, packed on demand. Choosing the point deliberately, rather than inheriting it, is what lets you hold inventory in its least costly and most flexible form while still quoting a lead time customers accept.

Forecast error does not disappear, it relocates

Building to stock puts forecast error into the warehouse, where it is visible, countable and slowly written down. Building to order puts the same uncertainty into capacity and material availability, where it is far less visible until a week arrives with more orders than hours. Plants that switch models and then plan as before are usually caught by this. The instruments change with the model: stock cover, ageing and write-down for one; order book coverage, promise-date reliability and queue length at the constraint for the other. Whichever you run, measure the thing that will actually hurt you rather than the thing your system already reports.

Mixed models need explicit rules, not judgement calls

Most plants of any size run both, because a small number of lines turn over predictably while the rest are sporadic. That is sensible, and it goes wrong when the classification is informal. Someone should own the rule that puts an item into one bucket or the other — demand frequency, variability, shelf life, value density, changeover cost — and someone should review it on a set cadence, because items migrate as they age. Without a written rule, stocked items accumulate through individual promises to individual customers, and nobody notices until the slow-moving stock report reads like the product catalogue.

Frequently asked questions

Can one plant run both models on the same line?

Yes, and it is common, but it requires a scheduling rule that says which work gets the slot when both compete. The usual approach protects customer-order work against promised dates and uses replenishment work to fill the gaps, which keeps the line loaded without pushing out a named delivery. The failure pattern is the reverse: stock builds get scheduled because they are easy to plan, and order work is squeezed into whatever remains until promise reliability collapses.

How do I decide which items to hold as finished stock?

Look at how often the item is ordered, how variable those orders are, what it costs to hold a unit, how long it stays saleable, and how much setup work a replenishment run consumes. Items ordered frequently with modest variability and low holding cost are the natural candidates. Items ordered rarely, or ordered in wildly different quantities, punish you twice — you hold the wrong amount and you hold it for a long time. Review the classification periodically, because demand patterns shift far faster than stocking policies do.

Does moving to make-to-order actually reduce inventory?

It reduces finished goods, which is the visible part, but it often increases raw material and component holdings because you must be able to start work when an order lands without waiting on the supply chain. The net position depends on the value profile: if most of your cost is added late in the process, the reduction is genuine; if most of the value is in a purchased component, you may simply be holding the same money in a different form and gaining flexibility rather than cash.