The short version: A near miss is a free warning. It is the thing that could have hurt someone and didn't, and it almost always turns up before the injury does. Near-miss, or good-catch, reporting is the best leading indicator most sites ignore. It doesn't die because workers stop caring. It dies from blame, from friction, and from reports that go nowhere.
Ask a crew what almost went wrong last week and you get a shrug. Ask what did go wrong and the stories pour out. That gap is the whole problem.
The thing that nearly hurt someone is the cheapest lesson your site will ever be handed. Most of the time nobody writes it down.
The cheapest warning you'll ever get
Herbert Heinrich claimed in 1931 that behind every serious injury sat 29 minor ones and 300 near misses. That 1-29-300 triangle still shows up in training rooms everywhere.
Take the exact numbers with a grain of salt. When Patrick Yorio and Susan Moore checked the idea against more than 27,000 workplaces and 668 deaths in Risk Analysis in 2017, they found no evidence for a fixed ratio. The pyramid is a story, not a law.
But the direction holds. Small warnings tend to run ahead of big ones, and they are nearly free to collect. The near miss is the one warning that shows up before the ambulance. That is what makes it worth chasing.
A leak that got reported, and nothing moved
On April 3, 2017, a steam pressure vessel at the Loy-Lange Box Company in St. Louis failed in a serious industrial accident. The US Chemical Safety Board investigated and issued a final report.
What stays with me is not the metallurgy. It is the timing. Employees had seen the vessel leaking in the days before it failed. A welder was called and couldn't get there. The vessel kept running as normal.
The warning was seen. It was reported. Nothing moved fast enough. A leak nobody acts on and a near miss nobody logs are the same thing wearing different clothes.
Now picture the opposite. The report reaches a supervisor the same shift. Someone makes the call to slow the line. The fix gets scheduled before the next run. That is what a system built to act on its own warnings looks like, and it is well within reach.
Why good people stop reporting
When a near-miss program goes quiet, it is almost never because workers stopped caring. It is because reporting stopped being worth it.
Blame kills it first. If the opening question is who messed up, people learn that silence is safer. Friction kills it next: a fifteen-field form on a computer nobody can reach from the floor does not get filled out.
The quietest killer is nothing happening. A worker who reports a hazard and watches it sit for a month has learned exactly what the report is worth. Do that twice and they are done. They are not being difficult. They are being rational.
The law is on the worker's side here, for what it's worth: firing or punishing someone for reporting an injury is illegal. But you don't build a reporting culture to satisfy a rule. You build it because it works.
What kills reporting, and what brings it back
| What kills reporting | What brings it back |
|---|---|
| Blame as the first response | No-blame framing: ask what the system allowed, not who slipped |
| A form nobody can reach from the floor | Reporting where the work is: a card, a text line, a code at the workstation |
| Reports that vanish into a folder | Fast closure: every report gets an answer, even if it is "not yet, and here's why" |
| Silence after something gets fixed | Visible action: post what changed because someone spoke up |
| Treating a rising report count as bad news | Treating more reports as a site that is paying attention |
| A "good catch" that earns a talking-to | Recognition: thank the person out loud, and mean it |
None of that needs software. A stack of index cards and a supervisor who reads them the same day beats an app that mails reports into a void. The tool is not the culture. The response is.
Start smaller than you think
The near miss you don't report is the incident you haven't had yet. You don't fix this with a policy launch and a poster. You fix it one report at a time, by making the first person who speaks up glad they did.
Ask the crew what almost went wrong this week. Write it down. Close it out where everyone can see. Then do it again next week.
The count will climb, and that is the point. A site catching fifty small things a month is reading its own warnings. A quiet site isn't safe. It's just quiet.
