Measurement and indicators

Critical to quality (CTQ)

In French: CTQ — caractéristique critique pour la qualité

A CTQ is the translation of a customer expectation into a measurable characteristic, with a target and limits beyond which the customer declares themselves dissatisfied.

Translating an expectation into a quantity

A customer expectation is rarely measurable as expressed. "A reliable delivery" cannot be recorded; "the percentage of orders delivered on the announced date" can be recorded every day. Going from the first wording to the second is the whole work covered by the acronym CTQ.

A complete characteristic carries three elements: the quantity measured, the target aimed for, and the limit beyond which the customer considers it no longer acceptable. Without a limit, the indicator allows no judgement — a value is observed without knowing whether it is good.

The limit comes from the customer, not from the current capability of the process. That is the most frequent inversion: what the process can do today is recorded and made the target, which guarantees an indicator always on green and a customer always unhappy.

From expectation to reading

The descent is made in steps. A general expectation breaks down into a few more precise requirements, each of which gives one or more measurable characteristics. "A responsive service" can thus break down into first-response time, resolution time, and number of contacts needed — three distinct quantities that a single indicator would have merged.

At each level, the question to ask is the same: how would I know, by looking at a measurement, that this requirement is met? As long as the answer remains qualitative, the descent is not finished.

The tree stops when each branch carries a quantity that someone can record without interpretation. A characteristic from which two people would draw two different values is not yet a characteristic: it is an appraisal, and it requires either a more precise definition, or a measurement system to be validated.

How many to retain

Few. A project that tracks about fifteen does not steer them: it publishes them. Three to five characteristics really tracked produce more effect than an exhaustive dashboard nobody reads in full.

Sorting is done on two criteria: the importance of the expectation for the customer, and the real margin for improvement. An important characteristic already held at a level the customer no longer notices is not a project subject — it is an asset not to be degraded, which is a matter of monitoring and not of improvement.

The characteristics retained become the project’s indicators, and they are the ones that will be recorded before and after. That choice therefore commits the demonstration of the result: an improvement measured on a quantity the customer does not perceive will convince nobody, even if the figures are exact.

The common mistakes

Setting the limit from current performance. The indicator stays green permanently and the project demonstrates there is nothing to do, when the customer was expressing themselves very clearly on the subject.

Stopping at a still qualitative wording. "Good file quality" is not a characteristic until what is counted has been said: rate of complete files at first submission, number of missing items per file, or something else — but something recordable.

Retaining too many. Beyond a few characteristics, none is really tracked any more, and the first difficult trade-off is settled by choosing the one that suits.

On the shop floor — IT services

The contractual commitment covered a response time, met more than 95 % of the time. Customers remained unhappy. Breaking down their expectation showed that what counted for them was not the response but the return to service, never measured: a ticket picked up in ten minutes then reopened three times counted as a success.

The other tools illustrated by this trade: services, it and engineering.

Frequently asked questions

What is the difference between the voice of the customer and a CTQ?

The voice of the customer is what was said, in the words of the person who said it. The CTQ is what was drawn from it: a recordable quantity, with a target and a limit. One is a material, the other a work product — and jumping from the first to the second without writing the intermediate step is the most common cause of indicators that speak to nobody.

How many CTQs should be retained for a project?

Three to five are enough in the vast majority of cases, and a single one is perfectly acceptable if it carries most of what is at stake. The number is not a sign of seriousness: a project that displays fifteen has generally avoided choosing, and its tracking will shrink by itself to the two or three that someone really looks at.

How do you define a CTQ on a qualitative need?

By looking for the observable trace of the need rather than trying to measure the need itself. "Feeling well received" cannot be measured, but the waiting time before the first contact, the proportion of requests handled without being redirected and the number of callbacks needed can all three be recorded, and their trend follows that of the feeling.

Are the limits of a CTQ the same as manufacturing tolerances?

Not necessarily, and confusing them is common. Manufacturing tolerances are often inherited from an old drawing or a machine capability, whereas the limits of a CTQ come from what the customer accepts. When the two differ, the gap is in itself information: either a precision nobody asks for is being produced, or deliveries fall outside what is suitable.