Where does invisible cost arise? The real price of scrap, downtime and rework

“Our scrap is three percent” is the most frequently heard and least useful sentence in production meetings. Knowing a plant's total scrap rate does not mean knowing where that scrap comes from. If the loss has no address, any reduction effort runs on guesswork.

Why a total rate is a poor indicator

A total scrap rate is an average, and averages hide differences. The same three percent can come from two very different pictures: small losses spread evenly across all lines, or a large loss concentrated at one station on one customer's job. The first calls for process improvement, the second for maintenance on a single die. A total rate cannot tell them apart.

To make loss addressable, the record has to hold three dimensions together: which work order, which station, which reason. Without all three, analysis turns into an argument whose outcome is known in advance — everyone defends their own line.

Scrap is not a rate but an address. A loss without an address shows up in the budget and cannot be found on the floor.

The four sources of loss

Invisible cost in manufacturing gathers in four items. All four behave differently; attacked with the same method, none of them falls.

• Setup scrap. Fixed per job and independent of quantity. On small batches it is the item that determines unit cost.
• Run scrap. Grows with quantity. Sensitive to material and machine condition.
• Downtime. Consumes no material but consumes capacity. Its price is the margin on the job that could not be produced.
• Rework. The most expensive item, because labour and energy are spent twice on the same product while the material counts once.

Separating setup scrap from run scrap

This separation takes one extra field and its effect goes straight to pricing. Setup scrap occurs the moment the job starts; the smaller the batch, the bigger its share of unit cost. Combined into one item, small batches are systematically underpriced.

The example below shows how the same product behaves at two batch sizes. Setup scrap is assumed at 120 units per job and run scrap at 1.5 percent of quantity; the figures are illustrative values chosen to show the method.

The table states what everyone on the floor knows intuitively but the system usually hides: small jobs are expensive. Without the separation, both jobs are priced at the plant average of two to three percent, and small batches close at a loss every time. In the costing layer this is why we keep setup and run scrap apart.

Costing downtime correctly

Measuring downtime is easy; costing it is hard. The common mistake is to book downtime only as lost hours. The right view asks what could have been produced in that hour: if the line is loaded, the price of downtime is the contribution margin of the lost output; if the line is idle, the price is only fixed cost. The same two-hour stop means very different amounts depending on order load.

The second issue is reason capture. Left to free text, downtime reasons become unanalysable; entries like “breakdown”, “no material” or “waiting” cannot be grouped afterwards. What is needed instead is a finite, plant-specific reason tree: at most two levels, at most six or seven options per level. The single most important indicator to watch is the share of “other” — if it is high, the tree does not cover reality.

Rework: the cost paid twice

Rework is the item most often hidden in the records, because the product does eventually ship and never appears as scrap. Yet labour, machine time and energy were spent on it twice. When it is not held as a separate item, that cost dissolves into overhead and which job was truly profitable becomes unknowable.

The correct record is plain: rework is opened as a separate operation linked to the original job. Two questions then become answerable — what this product really cost, and at which station the cause of rework arose. The second question is where improvement begins on the quality control side.

Seeing actual cost before the job ends

The most expensive property of invisible cost is that it becomes visible late. A loss learned at month-end is only accounting information; the window for intervention has closed. The same information seen when the job is thirty percent complete still allows a decision: revisit the setup, split the remaining quantity, discuss the delivery date with the customer.

The precondition is that cost is calculated as movements arrive, not at work-order close. As material issues, downtime, scrap and labour are posted, the work order's actual cost updates and the gap against plan becomes an indicator. We track this at work-order level on the the executive dashboard .

Cost learned after the job is a report; cost visible while the job runs is a decision.

Frequently asked questions

If our scrap rate is low, is this work necessary?

A low total rate does not mean the loss is evenly spread. In a plant at one percent, all of that loss may sit on one customer's job — a job actually produced at a loss. Without the breakdown, this stays invisible.

Does measuring setup scrap require extra hardware?

Usually not. Setup scrap follows from consumption between job start and the first conforming unit; if the work-order start and first quality approval are already recorded, the calculation can be made from existing data. Hardware improves precision — what makes measurement possible is the record itself.

What data is needed to cost downtime?

Three things suffice: the duration, the station's hourly capacity and the unit contribution margin of the product. Order-load information sharpens the calculation — a stop on a loaded line is lost margin, a stop on an idle line is only fixed cost.

Do these breakdowns add work for the operator?

Set up correctly, no. The setup/run split is derived from the work-order stage, so the operator enters nothing extra, and the downtime reason is a single-tap choice. Extra load appears when the reason list is kept long or free text is made mandatory.