Sampling in the Qualiopi audit: how the auditor picks the files they examine

A stubborn misconception has it that the Qualiopi auditor “checks everything”. That is materially impossible: the audit lasts a limited time, set contractually according to the organisation’s size and scope, while a year of activity represents dozens or hundreds of files. So the auditor does what every quality-system auditor does: they work by sampling. Understanding this mechanism completely changes how you prepare — and explains why organisations that cram three showcase files keep getting caught out.

Why the auditor samples

The Qualiopi audit assesses the compliance of your processes with the RNQ (Référentiel national qualité, the French national quality framework), not the perfection of each individual file. The logic runs in two stages:

1. The system: the auditor examines how you are organised — procedures, document templates, information flows, the improvement loop;
2. Its application: they then check, on a sample of real courses, that this system is actually applied.

This is a classic, long-theorised approach: a reference synthesis published by Elder, Akresh, Glover, Higgs and Liljegren in 2013 in Auditing: A Journal of Practice & Theory (“Audit Sampling Research: A Synthesis and Implications for Future Research”) recalls that every audit rests on sampling decisions — what to sample, how many items, how to interpret the anomalies found — precisely because exhaustive review is impossible (see the study). Transposed to Qualiopi: an anomaly found in a sampled file is never read as an isolated case, but as a signal about how the whole system works.

What the auditor actually samples

Without claiming to describe every certification body’s practice — each applies its own arrangements within its accreditation — sampling typically covers several crossed dimensions:

• Course files: complete files, from needs analysis to evaluation, chosen from delivered or ongoing courses;
• Categories of action: if you are certified for several categories (training, skills assessments, VAE — validation of prior experience —, apprenticeship), each must be covered by the audit — a multi-category organisation cannot hope that only its main category will be looked at;
• Delivery formats: classroom, distance learning, work-situation training… the auditor wants to see your system working in its different configurations;
• Funding channels: funded files and privately paid files can both be examined, since the framework’s requirements apply to the certified activity as a whole;
• Trainers: files delivered by employees as well as by subcontractors, since the subcontracting indicator is a control point in its own right;
• Sites: for a multi-site organisation, the certification body may sample the audited sites under the rules applicable to its accreditation, rather than visiting all of them at every audit.

The common thread across all these dimensions: the auditor chooses, generally from the list of your courses (session register, extract from your management tool) requested beforehand or consulted on site. You do not control the draw.

What this means for your preparation

The “any file” rule

The only robust strategy against sampling fits in one sentence: any file drawn at random must hold up. That does not mean every file must be perfect — the auditor assesses a system, and a one-off omission in an otherwise consistent file does not carry the same weight as a systemic absence. But it rules out “showcase” preparation: three immaculate files sitting among empty ones tell exactly the story you were trying to hide.

In practice:

• Standardise: the same templates (agreement, invitation, attendance records, evaluation, satisfaction survey) for every session, without exception;
• File as you go: a file is completed during the course, not the night before the audit — evidence carries dates, and a mass reconstruction shows;
• Unify the folder structure: one filing logic (by session or by indicator), known to the whole team, so that any piece of evidence surfaces in seconds on the day.

Deal with your blind spots before the auditor does

Go back through the sampling dimensions above and hunt for your weak points: the category of action you rarely deliver, the session run by a subcontractor, the distance course whose attendance tracking is shaky, the secondary site living its own life. These are precisely the configurations a good auditor will go after, because they test the limits of the system. A mock audit with a genuinely random draw — have someone else pick the files — is the best way to find those blind spots while there is still time to fix them.

On the day: play along with the draw

During the audit, resist the temptation to steer the auditor towards your best files: an experienced auditor senses it immediately, and it erodes trust for the rest of the day. Provide the requested list, let the draw happen, and put your energy into the clarity of your answers — knowing how to answer the auditor’s questions counts as much as the content of the files themselves. If a drawn file contains a gap, say so plainly and show how your system detects or corrects it: that is a far better answer than an improvised justification.

Sampling and non-conformities: how an anomaly is interpreted

When the auditor finds a gap in a sampled file, they try to qualify its scope: one-off or systemic? They may widen the sample on that specific point — drawing two or three additional files — to settle the question. A missing attendance sheet on an isolated session and a general absence of attendance evidence do not lead to the same conclusion: the first typically counts as a minor gap, the second goes to the very functioning of the system. This is also why hiding a known gap is counterproductive: far better to have identified it yourself, with a documented start of correction, than to let the auditor discover it and test its extent.

Take action

Sampling cannot be gamed — it can only be prepared for: a consistent system, evidence filed as you go and a random-draw mock audit will keep you clear of nasty surprises. To place this step within the whole journey — framework, costs, three-year cycle — see the complete Qualiopi certification page and download our free preparation guide.