The virtual classroom is the training format that produces the most data and the least evidence.
This sounds like a paradox, since every minute of a live online session is logged somewhere, by a platform that never forgets to switch itself on. It stops sounding like one the first time a funder, a corporate client or an auditor asks you, months later, to show that a given person attended a given session, and you discover that what the platform logged is not quite what you are being asked to show.
What the room proved without saying so
A sheet signed every half-day in a classroom links four things in a single gesture: a person, a moment, a place, and a trainer who saw the person sign. Nobody spells out these four links, because the paper and the walls make them for you. Whoever reads the sheet afterwards reads all four at once.
The virtual classroom removes the place and the paper. It removes neither the person, nor the moment, nor the trainer. But nothing links them in a single gesture any more, and that is where files come apart. The video-conferencing tool records a moment. The training centre knows the person. The trainer saw a name in a grid of thumbnails. Three true facts, held by three parties, in three files, and none of them is evidence until someone has brought them together.
What the tool records, and what it is worth
A video-conferencing tool produces, for each meeting, a participation report: a display name, a joining time, a leaving time, a duration. It is a useful document, and you should keep it. You should also know what it says.
It says that an account, or simply a browser if the meeting was opened by link, was connected to a meeting between two points in time. The display name is whatever the user typed or whatever their device carries: "Lea", "Lea's iPhone", "User 2", or the first name of a colleague who lent their computer. The duration is the gap between two technical events, not a measure of attention, nor even of physical presence in front of the screen. A connection of two hours and fifty-eight minutes in a three-hour session is consistent with full participation, and with a browser tab left open.
There is a simple way to state the limit. A connection log establishes that someone could reach the session and its material. It does not establish that they were taught. For a module followed alone, that gap is fatal, because nothing else in the file shows that a person was supported through it. A live session is in a better position, since the teaching is the session itself, with a trainer present in real time. But the reasoning applies word for word to the participation report: it proves a connection, and it proves nothing else.
Three pieces a live session needs, beyond the connection
Three things turn remote teaching into training that can be shown to have happened: that learners were supported by someone while they worked, that they were told beforehand what they were expected to do and for how long, and that their progress was checked along the way or at the end. A live session covers all three in practice. It rarely documents them.
The first is the best served and the least proven. The trainer is there, live, for the whole session. Who attests to it? Their own name in the participation report, with their joining and leaving times, and their signature on the session. A trainer missing from the attendance record is a trainer who cannot be shown to have taught, which is a familiar gap on paper sheets, moved onto a screen.
The second is settled before the session, by the invitation: date, times, link, stated length, trainer, and what is expected of the participant. It is an easy document to produce and is almost always missing from the file, because the link was sent in a message that nobody kept.
The third is the one that saves a file. A poll launched during the session, a short quiz at the end, an exercise posted in the chat: each produces a named, dated trace that the tool could not have invented and that "User 2" could not have submitted. Three questions asked at the forty-fifth minute do more for the evidence than three hours of connection.
The order of magnitude, to see where the time goes
Take a course of thirty-five hours, delivered in ten half-days of three and a half hours, for twelve learners. That makes one hundred and twenty attendances to establish. The video-conferencing tool produces ten reports, one per session, so about one hundred and twenty lines of joining and leaving times, under display names you did not choose, with the learners who reconnected after a dropout counted twice.
To turn that into evidence, someone has to match every line against a name on the enrolment list, decide who "User 2" was, add up the reconnections, add the trainer, and have the whole thing signed. At five minutes per session when all goes well, that is fifty minutes, call it an hour per course, provided it is done on the day. Six months later, when the funder or the client asks for any document showing that the course actually took place, the reports may well have been purged from the platform under retention periods that the vendor sets and that you do not know, and the matching can no longer be done.
A named remote sign-in, through a link sent to each participant at the start of the half-day, produces the one hundred and twenty attendances already attached to the right person and the right session, with the trainer's signature on each. The tool's report then becomes what it should be: a cross-check in case of doubt, not the evidence itself.
What is worth nothing, and turns up everywhere
A screenshot of the grid of faces is neither named nor verifiably timestamped, and it also raises a personal data question that no auditor asked you to raise. A sworn statement from the learner proves the session no better than the invoice does, and for the same reason: both are documents about the course, produced by the parties who benefit from it, not traces left by the course itself. A record "reconstructed" afterwards from the trainer's memory is a manufactured document, and it is worse than no document at all.
The same question, wherever you are
France, where the rules on attendance evidence for publicly funded training are among the strictest in Europe, has made remote training a particular point of scrutiny. Those rules apply only in France. Elsewhere the wording of contracts and audit checklists differs from one funder and one country to the next, but the question asked of a live remote session is the same everywhere: show that this person attended this session, with this trainer, for this length of time. A centre that can produce only a participation report under display names will struggle to claim it has checked anything at all.
The only test that counts
Pick a virtual classroom session from last month and, in five minutes, pull out the invitation with its times, the list of participants matched against the enrolment list, the joining and leaving times of each, the trainer with theirs, one named assessment trace, and check that all of it agrees with the hours shown on the certificate of completion. If the exercise takes an afternoon, it is not that the evidence is missing. It is that it was never assembled, and it will not assemble itself on the day someone asks for it.
Sources
This article cites no legal text. Its argument rests on what a participation report technically contains and on arithmetic set out in the body, which the reader can check for themselves.