Lean as a production model: choosing to run with less buffer on purpose

Choosing to operate without the protection stock used to provide

The defining commitment is to hold less of everything: less material between operations, less finished stock, less cushion in the schedule and less spare capacity kept for comfort. Inventory here is treated as a symptom rather than an asset, because each accumulation corresponds to a problem somebody decided to work around instead of solve. Reducing it does not by itself improve the plant, and that is where most attempts fail. Removing a buffer without repairing what the buffer concealed simply converts a cost into a stoppage. Sequence decides the outcome: make the process dependable, then withdraw the protection it no longer requires.

A supply base selected for dependability rather than quoted price

Operating with little material on hand transfers a great deal of responsibility upstream, so suppliers are chosen on consistency, delivery precision and willingness to work to a schedule rather than on the lowest number in a bid. That usually means fewer suppliers, longer relationships and more sharing of forward requirements than a transactional approach tolerates. Purchasing's measure of success moves from price variance toward delivered performance, including conformity at source, because there is no incoming stock to inspect your way through. The uncomfortable corollary is dependence: a smaller, deeply integrated supply base is both more efficient and harder to replace, and both things are true simultaneously.

Capital directed at flexibility, and what breaks when it is not

Money goes into making change inexpensive rather than into making runs long: faster setups, smaller and more numerous machines, equipment that can be relocated, and maintenance that keeps assets available instead of merely repaired. The characteristic failure is adopting the stock reduction without funding the capability that makes it survivable. A plant that halves its inventory while keeping lengthy changeovers and unreliable machines has simply become fragile, and the first supply disturbance proves it publicly. The second failure is running the model as a cost programme with a target attached, which delivers short-term savings, an exhausted workforce, and a quiet return to the old buffers.

Conformity built in, and the conditions where the model stops fitting

With no buffers, a defect halts the next operation almost immediately, so conformity has to be assured where the work is done rather than sorted afterwards. That is a demand on process capability and on the authority given to the people running it, not on inspection headcount. The approach extends well where demand repeats and stays reasonably level, since the whole arrangement depends on a predictable rhythm. It struggles where demand is genuinely lumpy, where upstream replenishment times are very long, or where the plant cannot decline sudden large orders. Under those conditions, deliberate buffering is the correct engineering answer rather than a lapse in discipline.

The demand shape it needs, and why the systems get lighter

Products with steady, repeating consumption fit naturally. Products ordered in unpredictable bursts do not, and forcing them into the same operating pattern produces a plant that looks efficient and delivers late. Many firms run a split, applying the model to their repetitive volume while managing the erratic tail on different terms. On systems the requirement is lighter than expected: fewer transactions, simpler triggers and exception reporting beat elaborate scheduling, because the floor is meant to respond to actual consumption rather than to a plan recalculated overnight. Heavy planning software in a plant like this often reveals that flow has not truly been achieved.

Frequently asked questions

Can a plant with volatile demand adopt this model?

Partially, and honestly rather than wholesale. The repetitive portion of the volume can run with minimal buffer, while genuinely erratic demand is handled with deliberate stock or reserved capacity that is planned rather than accidental. What does not work is applying uniform inventory targets across a mixed portfolio, because the erratic items will consume the flexibility the steady items depend on, and the plant ends up unreliable for both. Segment first, then set different rules for each segment.

What does this model ask of the finance function?

A different reading of inventory and of utilisation. Stock reduction releases cash once, and that is easy to applaud, but the ongoing changes are harder: machines deliberately not run flat out, smaller batches that raise apparent setup cost per unit, and capacity held in reserve. Traditional absorption measures punish all three. Unless the reporting is adjusted to reflect flow and delivered performance, the numbers will argue for exactly the behaviour the model exists to remove.

Why do inventory reductions so often reverse?

Because the stock was doing a job nobody replaced. Buffers absorb variability from unreliable equipment, inconsistent suppliers, quality problems and uneven demand. Taking them out without addressing those sources produces stoppages, and the organisation reasonably restores the protection. Reductions that hold are the ones that followed a measurable improvement in the underlying cause, which is why sustained progress tends to look slow and step-change targets imposed from above tend not to survive the first difficult quarter.