Near miss reporting is the cheapest source of safety information a firm has and the easiest to kill. It depends entirely on people telling you about things that did not hurt anybody, which they will stop doing the moment a report produces a conversation about blame or a form that takes ten minutes. This page is about the mechanics that keep the flow going, what a report should actually capture, and the step almost every scheme misses, which is doing something visible with them.
Make it fast, or it will not happen
The single biggest determinant of reporting volume is how long a report takes. Anything over a minute competes with the job somebody was doing, and loses. What you need is short: what happened, where, when, what could have happened, and who is reporting it if they are willing to say. Anything else can be added by whoever reviews it. Schemes that ask for a risk rating and a causal category at the point of capture collect very few reports and very good data about the three people who fill them in.
Reports go down when people are afraid, and that looks like improvement
A falling report count is ambiguous and firms consistently read it the wrong way. If your numbers dropped last quarter, the first hypothesis should be that people stopped telling you, not that the workplace improved. The test is easy: ask a few of them. Consequence-free reporting is a management commitment rather than a software feature, and no product can substitute for it, though a scheme that takes ten seconds removes one of the excuses.
What a near miss incident report should capture
The description in the reporter's own words, because paraphrase loses the detail that matters. Location and time. The potential outcome, honestly stated, since that is what decides how hard to look. What already in place limited the outcome, which is often the most interesting field and almost never asked. Then, added by the reviewer: cause, action, owner, date, and whether an assessment now needs revising.
Close the loop visibly, or the scheme dies quietly
People report when they see reports change something. Publishing a short note of what was reported and what was done about it, monthly, does more for reporting rates than any campaign. It also forces the discipline of assigning actions, and it exposes the pattern nobody notices in individual reports: four instances of the same near miss in a quarter is a control that is not working, which is exactly what the free risk assessment form worksheet on this site treats as an out-of-cycle revision.
Questions people ask about near miss reporting
What is a near miss incident?
An unplanned event that could have caused injury, ill health or damage but did not. Some firms distinguish near misses from unsafe conditions or undesired circumstances; the practical rule is the same, which is to capture it and look at what it reveals about a control.
Can you give near miss examples worth reporting?
A load slipping from a forklift with nobody underneath. A vehicle reversing to within a metre of a pedestrian. A tool dropped from height inside an empty exclusion zone. A wrong chemical decanted and spotted before use. Each one is a control that failed at no cost.
Should near miss reports be anonymous?
Allow anonymity, do not require it. Named reports are more useful because you can go back and ask, and most people will give their name in a culture where doing so is safe. Requiring names in a culture where it is not simply moves the reporting to the pub.