After an incident: getting the finding to every exposed worker

Something happens. A near miss in the pick aisle, a hand caught during a changeover, a fall from a loading dock edge. Within days the procedure is tightened, prevention is consulted and the investigation produces findings. Then comes the part that quietly fails. The site where it happened talks about it. Comparable sites hear nothing, or a version of it in a mail. Six months later the same gap opens somewhere else, and nobody can show who was retrained, when, on which version.

Why the finding stays where it happened

Nobody decides to keep a finding local. It stays local because of how the follow-up is organised. The investigation is run by the site, so the corrective measure is written for the site. The conversation happens at that site's next toolbox, in that site's language, on the crew that happened to be in. The other sites get a summary in a mail, which is read by whoever reads mails, and the night crew at site two learns about it when something similar happens to them.

The law is clear about where this should end. Belgian Codex art. I.2-21 requires training to be repeated and adapted as risks evolve, and names new equipment and new technology as triggers in their own right. An incident is the plainest evidence that a risk was not controlled. Art. I.2-20 adds that only workers with appropriate instructions may enter zones with serious and specific hazards, which makes the retraining an access condition for those zones. In the Netherlands, Arbowet art. 8 requires instruction that is effective and adapted to each worker's tasks, and lid 4 requires the employer to supervise that it is followed. VCA question 4.1 names findings from incident investigation as required toolbox content, and the point of that rule is the sites where it has not happened yet.

Who is exposed

Scoping decides whether the retraining lands. Send it to the whole company and people learn that safety updates are noise. Send it to the one site and you leave the same exposure open elsewhere.

That last line is not a detail. In the Dutch accident monitor for 2024, agency workers were the victim in 18 percent of investigated accidents while making up about 4 percent of the working population. They are also the people most often missing from the list after an incident.

Test the decision, not the rule

The retraining after an incident is not a reminder that a rule exists. Most people involved in an incident knew the rule. What failed was the judgement at the moment of decision: the guard that was usually fine to open, the edge that did not look like an edge in that light, the reversing vehicle nobody expected at that hour.

So the check should ask about that moment. Describe the situation as it was: the machine jammed, the line is behind, the technician is at another site. Ask what the worker does next, and why the tempting option is wrong. A scenario like that, answered per person, tells you whether the finding changed anyone's behaviour. A signature under a mail does not. It also connects to the check the worker makes on the spot: the LMRA is where the finding has to show up in practice.

Record it against the new version

The record that closes the loop has four fields per person: who, on which date, on which version of the procedure, and with what result on the check. The version matters most. A worker who completed the old procedure in March and the tightened one in October is in a different position from one who has only the March record, and your record should show the difference without anyone reconstructing it.

What the accident data says about priorities

The figures are as of September 2026, for the year 2024. The Nederlandse Arbeidsinspectie's Monitor arbeidsongevallen 2024 counts 1,990 reportable accidents with 2,001 victims, 58 of them fatal. Falls account for 35 percent, contact with a work implement or object for 31 percent, being struck by an object for 17 percent and vehicles or mobile equipment for 8 percent. In Belgium, Fedris recorded 97,591 workplace accidents in the private sector in 2024.

Read those against a typical follow-up. The two largest categories, falls and contact with equipment, are specific to this edge, this machine, this changeover. They are not addressed by a general session on being careful, and they are exactly the findings that repeat at a comparable site when they stay local.

How Aristotl handles it

Aristotl turns the investigation report and the tightened procedure into a course of 3-5 minutes that workers complete on their own phone, in their own language, with a knowledge check that asks what they would do. Your team reviews it before it is published. The course is assigned per site or team, so it goes to the comparable sites and crews rather than to the whole company, and temps join through a link or QR code without a company email. The finding itself can go out as a must-acknowledge update with named receipts per person.

Completion is viewable per site and version in the platform, so a completion on the old procedure shows as exactly that. The practical follow-up on the floor, the investigation itself and the supervision stay with your prevention advisor and line managers.

Take your last incident investigation and list the sites and crews with the same exposure. Then check how many of those people you could show, by name, were retrained on the new version.