Measurement and indicators

Quality metrics

In French: Indicateurs qualité

Scrap rate, rework rate, first-pass yield, cost of poor quality: quality metrics put a figure on what defects cost, and above all on where they occur.

The families of metrics

Those that count units. Scrap rate, rework rate, conformity rate. Simple, immediately understood, they have one flaw: a part reworked three times counts as one, and the redone work disappears from the board.

Those that count defects. Since a single unit can carry several defects, these metrics measure the real load of poor quality. They are the necessary step towards the sigma level.

Those that look at the whole process. First-pass yield — the proportion of units that go through every step without any rework — is the harshest and the most instructive: it refuses to offset a mediocre step with an excellent one.

Those that count in euros. The cost of poor quality adds up scrap, rework, inspection, delays, complaints. It is the only language that makes a management team decide, and the only one that makes defects of different natures comparable.

First-pass yield, and why it surprises

A ten-step process at 99 % conformity per step does not deliver 99 % of good units first time: it delivers about 90 %. Yields multiply, they do not average.

It is this calculation that makes visible what no step could see on its own. Each step owner shows a reassuring figure, and the complete process loses one unit in ten — which goes back to rework, consumes lead time and work, and appears nowhere.

This finding is often the trigger of a project: it turns ten good local results into one bad overall result that nobody could ignore.

The cost of poor quality

It breaks down into four parts, and their proportion is a diagnosis in itself. External failures — what the customer suffers: complaints, returns, warranties. Internal failures — scrap and rework detected before shipment. Appraisal — inspections, tests, audits. Prevention — training, error-proofing devices, robust design.

An organisation whose cost is mostly in appraisal sorts a lot and prevents little: it pays to find the defects instead of paying not to produce them. Moving the spend towards prevention is, over time, the best return available.

The external share deserves particular attention: it is almost always underestimated, because the loss of a customer’s trust appears on no accounting line.

The common mistakes

Tracking a scrap rate without tracking rework. An organisation can show a scrap rate close to zero by reworking massively — the cost is there, the delay too, and the indicator shows nothing.

Adding up defects of different severity. A scratch and a functional non-conformity each count as one, and the total no longer means anything. Either you weight, or you track the families separately.

Multiplying the metrics. A board that carries fifteen guides no decision; three well chosen ones do.

On the shop floor — Aerospace supply

A workshop tracked a conformity rate of 98.5 % per step and was satisfied with it. Calculating the first-pass yield over the seven steps gave 90 %: one part in ten went through at least one rework. The cost of that rework exceeded the cost of scrap by a factor of four.

The other tools illustrated by this trade: electronics and aerospace.

Frequently asked questions

What is the difference between scrap rate and first-pass yield?

The scrap rate counts what is thrown away; first-pass yield counts what goes through the process without any rework. A part reworked then accepted weighs nothing in the first and degrades the second. That is why the second is the best revealer of hidden work.

How do you put a figure on the cost of poor quality?

By adding up what would not have been spent if everything had been right first time: scrapped material, rework hours, inspection hours, extra transport, penalties, complaint handling. The difficulty is not the calculation but access to the data, often spread between production, quality and accounting.

Should a quality metric be displayed to the team?

Only if the team has a grip on it. A scrap rate that depends on a parameter set elsewhere turns into a repeated reproach, and makes the display lose all credibility for the metrics on which the team can really act. The selection criterion is therefore the power to act, before technical relevance.

Should zero defects be the aim?

As a direction, yes; as a universal numerical target, no. The level aimed for follows from the stakes: a safety component and a cosmetic part do not call for the same effort. Setting the same target everywhere leads to over-investing where it does not count and under-investing where it does.