Your learners log in at 6 a.m., on a Sunday, in twenty-minute chunks. And your Qualiopi auditor is asking you for „attendance sheets". The two seem incompatible, and plenty of training providers end up cobbling together unverifiable spreadsheets or getting backdated PDFs signed. That is exactly what triggers a non-conformity.

Good news: for asynchronous distance learning, the French Labour Code does not require a signature per half-day. It requires an individualised training protocol and evidence that the action actually took place. That difference completely changes how you organise your day-to-day.

Here you'll find the four methods actually used to produce digital attendance records that hold up in front of an auditor, from the most automated to the most manual. For each one: what it really proves, what it costs you in time, and where it falls short. And at the end, the precise list of documents to file in your evidence folder, session by session.

What Qualiopi really requires in e-learning (and what it doesn't)

In open and distance learning, the expected proof is not a signature at a fixed time: it's a body of traces showing that the learner followed and completed the training.

The framework rests on two texts. First, article D6313-3-1 of the French Labour Code on Légifrance, which states that a distance learning action takes place according to an individualised protocol covering three elements: the technical and educational support made available to the learner, the information about the learning activities and their average duration, and the assessments that punctuate or conclude the action.

Then the national quality framework, set by decree no. 2019-564 of 6 June 2019: 7 criteria, 32 indicators, mandatory since 1 January 2022 for any provider that wants access to public and pooled funding (OPCO, CPF, France Travail, regions, local authorities).

In practice, the auditor is trying to answer three questions:

  • Did the learner actually access the content, and over what period?
  • Did they produce something (assignments, quizzes, assessments) that proves their activity?
  • Did they have access to support, and did that support leave a trace?

What the auditor does not ask for: a handwritten signature per half-day for asynchronous sequences. That would be a fiction, and a well-documented fiction is still a fiction.

Two nuances to watch. As soon as a sequence is synchronous (virtual classroom, video conference, in-person), hourly attendance records become the norm again, and they're almost always required. And above all, your funder can be stricter than the law: the terms and conditions of an OPCO or a regional agreement take precedence over your convenience. Read them before choosing your method. If you want the details indicator by indicator, the article on what auditors really expect when it comes to Qualiopi evidence pairs well with what follows.

Method 1: a platform that produces the evidence for you

The shortest path is to use an e-learning platform that natively timestamps every login, every completed module and every assessment, then exports the whole thing as one document per learner.

That's the principle behind Skilzy, a French platform for learning and using AI: more than 15 programmes (image creation, UGC videos, ads, faceless channels, avatars and voice-overs, music, automation with n8n, prompt writing, AI-assisted development). A built-in „AI Lab" lets learners use the real tools directly inside the platform, with credits included, so they don't have to stack up subscriptions of their own.

For a training provider, the benefit is twofold. The catalogue is ready to offer, and two certifications registered in the Répertoire spécifique are state-recognised and eligible for funding: the RS7439 certification in AI content marketing and RS6792 in AI and sales. On the compliance side, learner activity is tracked by design: login dates and durations, modules opened and completed, quiz and assessment results, work produced in the Lab, support exchanges.

These elements feed straight into the file an auditor expects. The compliance offer for training providers sets out the available exports and how to attach them to your agreements.

Two honest caveats. The platform supplies the raw material: building the file, keeping it consistent with the quote, the agreement and the completion certificate remains your responsibility, and nobody can carry that for you. And if you're using several tools in parallel, you'll need to consolidate the exports.

On the testing side, the discovery demo gives you 7 days of access with no credit card, including 1 image, 1 video, 1 music track and 10 messages: enough to check the learner experience before committing to anything. The consumer plan starts at €29.90 per month with no commitment, and a dedicated offer exists for training providers. If you're still weighing up your technical foundation, first compare the options in the guide to choosing an e-learning platform for your training organisation.

Method 2: your current LMS plus an attendance module

If you already host your content on an LMS like Moodle, Chamilo, LearnDash or 360Learning, the evidence often already exists in the activity logs: the work is making it readable.

An LMS records logins, time spent per module and assessment results. SCORM or xAPI (also called Tin Can) content additionally reports completion status and score. Many tools offer an attendance module or plugin that turns these logs into a dated, electronically signed record.

Two settings not to overlook. Connection time is not training time: a tab left open for three hours manufactures false data, and an attentive auditor will notice. Configure an automatic logout after inactivity, for example 15 or 30 minutes. Then export to timestamped PDF rather than editable CSV: a raw spreadsheet file can be tweaked in two clicks, which weakens its evidential value.

Main limitation: raw logs are unreadable for someone discovering your tool for the first time. Plan a formatting template, one page per learner, with your organisation's name, the title of the training action and the session dates.

Method 3: a dedicated electronic attendance tool

Specialised solutions (Edusign, Digiforma, Dendreo, Ypareo and equivalents) bring what an LMS does badly: named, timestamped signatures, especially for synchronous sequences.

The principle is simple: the learner receives a link or a code by email or SMS, signs from their phone, and the tool generates a consolidated sheet per session. The signature relies on the European eIDAS regulation, which distinguishes three levels (simple, advanced, qualified). For attendance records, a simple electronic signature is accepted in practice as long as you can demonstrate the signer's identity, the exact date and the integrity of the document. The qualified level, far more expensive, isn't expected here.

This is the best option for virtual classrooms, in-person gatherings and blended sessions. Pricing varies by vendor (per user, per session or as an annual package): check the current rate card, it moves often.

Limitation: this tool proves presence at a given moment, not asynchronous engagement. If your programmes are mostly self-paced, it doesn't replace your platform's activity records, it complements them. And without a connector to your LMS, you'll be re-entering sessions by hand.

Method 4: the hand-signed PDF, the minimum viable option

Sending an attendance sheet as a PDF, having it printed, signed, scanned and sent back is still possible, but it's the most fragile and most time-consuming method.

It holds up in one case: low volume, a few sessions a month, with clearly dated synchronous sequences. Beyond that, you'll spend a huge amount of time chasing learners, and the return rate drops fast.

The risks are real. A scanned signature is worth no more than a piece of paper: nothing indicates the actual date of signature, which opens the door to suspicions of backdating. Sheets signed after the fact all at once, all in the same ink and on the same day, are a classic red flag in an audit. And an unprotected PDF is easy to modify.

If you stick with this method, lock down at least three things: one sheet per session and per date, not a monthly summary sheet; a saved email trail for sending and returning (the message timestamp becomes your proof of date); and systematically named archiving.

Quick comparison of the four methods

Method What it proves Effort per session Suited to
Integrated platform (Skilzy) Logins, progress, assessments, work produced Almost none, automatic export Asynchronous programmes and high volumes
LMS plus attendance module Logins and completion, signature depending on plugin Low once configured Providers already running an LMS
Dedicated attendance tool Named, timestamped presence Low, automated sending Virtual classrooms and synchronous sessions
Hand-signed PDF Declared presence, uncertain date High, manual chasing Low volume, occasional stopgap

The evidence file: the checklist to build for each learner

A solid file contains seven items per learner, all dated, consistent with each other and unambiguously traceable to the same person and the same training action.

  1. The individualised training protocol, handed over before the start, with the activities, their average duration, the support arrangements and the planned assessments.
  2. The timestamped activity record: login dates, durations, modules opened and completed.
  3. Assessment results: entry positioning, intermediate quizzes, final assessment.
  4. The learner's output: assignments, deliverables, exercises, screenshots of work done in a lab or an integrated tool.
  5. Traces of support: emails, in-platform messages, tickets, meeting notes. This is the most frequently forgotten item, and one of the most closely examined.
  6. Signed attendance sheets for every synchronous sequence, one sheet per date.
  7. The completion certificate sent to the funder, plus the end-of-training attestation given to the learner.

A few rules that keep you out of trouble. Name your files following a single scheme, for example 2026-03-12_LASTNAME_Firstname_activity-record.pdf: an auditor who can find a document on their own in ten seconds is a reassured auditor. Check that dates are consistent across the agreement, the activity record and the completion certificate, because date inconsistencies are the most common cause of non-conformity. Keep everything for at least three years, the length of the certification cycle, bearing in mind that an administrative check can cover earlier periods; many providers keep five years, with the retention period written into their GDPR register.

The three mistakes that cost you dearly: records generated en masse the day before the audit, connection times identical to the advertised durations (nobody completes a 45-minute module in exactly 45 minutes), and an individualised protocol sent after the training has started. On that last point, if you build your offer from external content, check that the ready-made AI training catalogue you're using does provide average durations per activity: without them, your protocol is incomplete.

Where to start this week

Take a session that recently finished and try to reconstruct the seven items. Whatever you can't find in under ten minutes is exactly what will be missing on audit day. Then choose your method based on how much of your programmes are asynchronous: an integrated platform if everything happens self-paced, a signature tool as a complement as soon as you schedule virtual classrooms. The rest is a matter of habit, not technology.