Make-to-stock: producing ahead of demand and living with the forecast

The bet a plant makes when it builds before the order arrives

Make-to-stock commits the factory to a production plan built from a demand signal that has not happened yet. Scheduling is decoupled from order intake, so the line runs to a replenishment target rather than to a named customer promise. That works where demand is broad, repeated and reasonably stable across many buyers, and where customers compare on availability rather than on specification. It stops working the moment product variety multiplies, or when the sales pattern is driven by a handful of large accounts whose reorder timing you cannot see. The clearest test is simple: can you name the product before anyone orders it?

Where the stock piles up, and which pile hurts most

Three pools behave differently. Raw material is comparatively forgiving, because it can usually be redirected to another product. Work in progress stays small, since the line runs continuously toward a known specification. Finished goods absorb everything the model gets wrong, and that pool is the expensive one: it carries the full converted cost, it occupies conditioned space, and it ages. Obsolescence rarely arrives as a single write-off. It accumulates as slow-moving lines nobody wants to declare dead. Plants that survive on this model treat the ageing profile of finished goods as an operational measure owned by production, not a finance report reviewed after the quarter closes.

Buying long while selling into an uncertain week

Suppliers are contracted against the same forecast that drives the line, which means purchasing commits to volume before revenue is confirmed. That buys genuine advantages: negotiated pricing, scheduled deliveries, materials arriving in economic quantities rather than urgent ones. It also extends the exposure backwards, because a demand drop leaves you holding raw material bought for products you will now not sell. Procurement earns its keep here by writing flexibility into the commitment: volume bands instead of fixed quantities, call-off arrangements against a blanket agreement, and a deliberate look at which components are generic enough to be worth stocking against several finished lines at once.

Equipment chosen for repetition, and the volume it needs to justify itself

Assets are specified for repetition rather than range. Changeovers are infrequent, so tooling can be dedicated, material handling can be fixed, and automation is easy to justify because the same motion repeats indefinitely. That is a heavier upfront commitment than a flexible cell, and it is only defensible if the product outlives the payback on the machine. Growth is straightforward while you are growing the same thing: add a shift, add a parallel line. It stalls when variety enters. Each new variant consumes changeover time on equipment chosen precisely because it does not change, and the plant discovers its usable capacity was never what the nameplate implied.

Forecast error, and the systems that either catch it or bury it

The characteristic failure is not a production failure at all: the plant hits its plan and the market does not turn up. The mirror image is just as common, where a promotion or a competitor outage empties the shelf and the line cannot respond because it was scheduled weeks out. Both are visible early in the right system. This model needs planning software that carries a forecast, nets it against on-hand and open supply, and shows the resulting schedule in a form a planner can override deliberately. What it must not do is quietly reforecast from its own shipment history, which turns every stockout into a permanent cut in future supply.

Frequently asked questions

How much finished goods cover should a make-to-stock plant hold?

There is no universal answer, and any figure quoted without your own data is noise. The working method is to size cover against two things you can measure: how variable demand is for that specific line, and how long it takes you to replace it once you decide to. A predictable product on a short production cycle needs very little. An erratic one made on a congested line needs far more. Do the calculation product by product, because category averages hide the lines that actually cause the stockouts.

Can one plant run make-to-stock and make-to-order side by side?

Routinely, and most sizeable plants do. The usual split is by volume and predictability: high-runners are built to stock so they ship from the shelf, while the long tail is built against firm orders. The difficulty is scheduling, because replenishment work is infinitely deferrable and a customer order is not, so the tail quietly gets pushed back. Protect it with reserved capacity, meaning a defined block of line time that stock replenishment is not permitted to consume, rather than trusting a planner to hold that line under pressure.

What warns you that a make-to-stock model is failing before the write-off lands?

Composition, not value. Total inventory can look flat while what sits inside it rots, so track the share of finished goods that has not moved within its normal cycle, and watch that share line by line. Two other signals matter. Rising changeover frequency says variety has outgrown the equipment. A widening gap between forecast and actual at item level, even when the aggregate looks close, says offsetting errors are hiding on the shelf and each one is a real customer problem.