Criterion 7: satisfaction and complaints, listening to improve

The criterion that proves all the others

Criterion 7 of the Référentiel national qualité closes the standard with a sober formula: "The provider collects feedback and handles complaints." Only three indicators, out of the 32 in the RNQ: 30 (collecting feedback), 31 (handling complaints) and 32 (continuous improvement). Few indicators, a role apart. It is the loop: listen, analyze, act.

Its position at the end of the journey is misleading. It is not the last box to tick, it is the only proof that the six preceding criteria truly produce value. It turns a documentary system into a living one. A provider can line up flawless onboarding procedures, detailed programs, exemplary trainer follow-up, and still be an administrative shell if nothing comes back from the field to adjust it. Criterion 7 is the point where the auditor checks that your approach is still moving.

The reference framework here is the Guide de lecture Qualiopi in version 9, published by the ministère du Travail. In force since 8 January 2024 and enforceable since 8 March 2024, it sets, for each indicator, the expected level, the admissible evidence and the type of non-conformity incurred. It is the document your auditor has in front of them. Keep it in front of yours.

Indicator 30: gather every voice, not only those who reply

The most common mistake: surveying only the learners. Indicator 30 demands more. You must solicit ALL stakeholders. The beneficiaries, of course, but also the teaching teams and trainers, the employing companies and the funders (the latter at least once a year). Overlook one audience, and it is a non-conformity, no matter how refined your questionnaire.

Feedback is gathered at key moments, never at a single point. During the service, at the end of the service (the immediate evaluation, which captures the fresh reaction) and afterward (the delayed evaluation). This last one is too often neglected. It measures 3 months later for technical training, up to 6 months for behavioral training, whether the skills hold up once back on the job. It is also your best sales argument: proof of real transfer weighs far more than immediate satisfaction, and it feeds directly into your improvement plans.

Yet sending the survey is not enough. The indicator calls for a system to follow up with non-respondents (2 to 3 reminders, the range that maximizes the response rate) and above all the traceability of the requests: date, channel, response rate, reminders sent. A questionnaire sent out with no reminder and no record is a failing, even if its content is beyond reproach.

Indicator 31: the most binary point in the standard

Here is the indicator that sinks otherwise watertight files. It requires a formalized AND published procedure for handling difficulties, complaints and unforeseen events: an accessible collection channel, an acknowledgment of receipt, an entry in the register, a traced response, a closure of the case. Weigh the word published. A procedure buried in a binder that no one knows about does not exist in the eyes of the audit.

Two confusions cost the certification. The first: mistaking a negative remark in a satisfaction questionnaire for a complaint. No. Satisfaction (indicator 30) and complaints (indicator 31) follow two distinct logics. One is statistical and recurring, the other is individual and traced from end to end. A single form cannot carry both flows. The second: believing that an oral response is enough. Without a written record (acknowledgment, register, response, closure), the complaint is deemed unhandled. Traceability is the proof, not the gesture.

And above all, engrave this: indicator 31 tolerates no minor non-conformity. The slightest failing is MAJOR and blocks the certification. It can still be resolved within 3 months, but it is the point to settle as an absolute priority before any audit.

The first provider has a perfect file on paper and fails. The second has a real problem and succeeds. That is the whole difference between the formality and the engine.

The traced loop: reading a register like an auditor

A complaints register reads like a closed loop: date, origin, nature, handling, response to the complainant, closure date. A register that lines up only open entries does not prove handling, it proves its absence. The auditor is not looking for a well-filled register, they are looking for cases that close.

Now for the virtuous counter-example. A provider receives a complaint about response times judged too long. It acknowledges receipt within 48 hours, enters the case in the register, investigates, replies to the complainant in under two weeks, then overhauls its administrative organization. It has just turned an irritant into evidence. This is exactly what indicator 32 expects: an explicit causal link between a piece of data from 30 or 31 and a corrective action. We received this feedback, so we changed this. An action plan with no identified data source looks cosmetic.

The weakness is almost never a lack of quality, it is a lack of proof. A provider can handle its complaints to perfection and fail for want of having traced them. The rule fits in one sentence: formalize before doing well, then prove that you do.