Method

Voice of the customer

In French: Voix du client (VOC)

The voice of the customer is the set of expectations expressed by those the process serves, collected in their words before being translated into measurable characteristics.

Collect before translating

The voice of the customer refers to the raw material: what customers say, demand, work around or give up on. It is written in their vocabulary, including when that vocabulary is vague — "it takes too long", "you never know where it stands", "you always have to call back".

The temptation is to translate immediately into a numerical objective. That is a mistake of order: rewording "you never know where it stands" as "five-day lead time" replaces a request for visibility with a promise of speed, and the customer will remain dissatisfied even if the lead time is met.

Collection therefore precedes interpretation, and it is kept as it is. The original wordings serve as a safeguard when, three months later, the project team discusses a trade-off: one goes back to them to check that there has been no drift towards what was convenient to measure.

Where to find it

The sources that already exist. Complaints, call reasons, credit notes, returns, comments, tickets. They have the advantage of being free and the flaw of being biased towards loud dissatisfaction: those who leave without saying anything are not in them.

Interviews. Expensive but irreplaceable for understanding the why behind a request. A few well-conducted interviews teach more than a widely distributed questionnaire, because they make it possible to follow up on what had not been anticipated.

Observation. Watching a customer actually use the service reveals the workarounds they never think to mention, because for them they have become a habit and not a problem.

Lost customers. The most instructive and least solicited source. Those who left have a precise reason, and they often give it willingly if asked without trying to win them back in the same conversation.

Sorting what has been collected

Not everything that comes up carries the same weight. An expectation cited by one customer in thirty is not worth an expectation cited by one customer in three, and sorting means counting — which requires having collected enough material for the count to make sense.

Nor are expectations all of the same nature. Some are taken for granted and are noticed only when missing; others are explicitly requested; others still are never voiced but delight when present. Confusing those three categories leads to investing in what will never make a difference.

The next step is the translation into critical-to-quality characteristics: that is where "too long" becomes a measurable quantity, with a target and a limit. That translation is a piece of work in its own right, and it is done after collection, never during.

The common mistakes

Sending out a questionnaire instead of collecting. A questionnaire returns only the answers to the questions asked: it confirms hypotheses instead of revealing them, and systematically misses what nobody had thought of.

Asking customer service instead of the customers. Front-line teams have valuable knowledge, but filtered by what they can handle — the requests to which they never have an answer end up no longer being reported to them.

Rewording the verbatims to make them presentable. The customer’s exact language is what makes it possible to decide later between two interpretations; polished and summarised, it serves only to illustrate a conclusion already written.

On the shop floor — Health insurance

A project aimed to reduce the reimbursement time, cited in almost every complaint. The interviews showed that the irritation came mainly from the lack of information during the wait: several policyholders said they would accept a longer time if they knew where their claim stood. The project changed its object and the tracking indicator with it.

The other tools illustrated by this trade: healthcare and life sciences.

Frequently asked questions

How many customers need to be interviewed?

Fewer than one thinks to identify the themes: interviews stop bringing anything new after about ten, provided different profiles are covered. On the other hand, ranking those themes requires counting on a larger volume, often from existing sources such as complaints.

How do you collect the voice of the customer on an internal process?

By identifying the customer in the process sense, that is, the one who receives the output and depends on it to work. The accounts department is the customer of the sales department for the transmission of orders, and its expectations are collected in exactly the same way: interviews, observation, counting of rework.

What should be done with contradictory expectations between customers?

Treat them as a segmentation signal rather than as a trade-off to make. Opposing expectations almost always come from two different populations with different uses; seeking the compromise produces an offer that satisfies neither, whereas distinguishing them makes it possible to serve each correctly.

Does the voice of the customer replace satisfaction indicators?

No, the two answer different questions. A satisfaction indicator says that a level has deteriorated; it says neither why nor what to act on. Qualitative collection gives the cause, but does not make it possible to follow a trend over time. Organisations that only measure know they have a problem without ever knowing which one.