A Near-Miss Happened but Nothing Broke: Decision Tree

Why this matters

A near-miss is a free lesson, and most shops throw it away. Something almost went wrong - the ladder shifted but nobody fell, the wrong valve opened but got caught before it mattered, the invoice almost went out with a whole section missing - and because nothing actually broke, the natural move is to exhale and get back to work. That exhale is the mistake. A near-miss (any event that could have caused injury, damage, or a failed job but did not, by luck or a last-second catch) is the same underlying defect as the real failure, minus the bill. This tree is what you do in the minutes and hours after one so the warning does not go to waste.

Start here: is anyone or anything still in danger

Before you analyze a thing, make the scene safe. A near-miss often means the hazard is still live.

• If the event involved live electricity, gas, water near power, height, stored pressure, or a moving machine, neutralize that hazard first: de-energize and verify dead, close the valve, relieve the pressure, secure the ladder or the load, move people back. Confirm it is actually off, do not assume.
• If someone got rattled, check on the person before the process. A shaken tech should not climb back up for a minute.
• Only once the immediate danger is gone do you move to capturing and diagnosing.

Diagnosing a hazard that is still armed can wait sixty seconds. Making it safe cannot.

Then: capture it before the memory fades

The details that tell you the real cause evaporate within the hour. Lock them down now, clean them up later.

• Write or voice-note the plain sequence: what was supposed to happen, what actually happened, and the exact point luck or a catch saved it.
• Photograph the setup before anyone tidies it. The leaning ladder, the open panel, the draft invoice on screen.
• Note the conditions: end of a long day, a rushed schedule, a new hire, an unfamiliar site, a tool that was not on the truck. These are your real suspects.

Was this luck, or was the system going to allow it anyway

This is the branch that matters. Ask: if we ran that same job the same way ten more times, how many would end badly?

• If the only thing that saved it was luck or one person's fast hands - the system was going to let the failure through. Treat it exactly as if it had happened. Same investigation, same fix. The absence of damage is not evidence you are safe; it is evidence you got away with it once.
• If a real control caught it - a checklist step, a second set of eyes, a guard, a verification - then the system worked as designed. Confirm the control is not fragile (one distracted person from failing) and move on. Do not over-engineer a save that already works.

When in doubt, treat it as the first case. Under-reacting to a near-miss is the far more expensive error.

One-off or the first visible instance of a pattern

Now decide how far the fix has to reach.

• If this near-miss shares conditions with past events - same task, same time of day, same tool gap, same step skipped - it is not a one-off. You are seeing the visible tip of a recurring system problem. The fix has to change the system, not the person.
• If it is genuinely isolated - a freak alignment of circumstances unlikely to repeat - a targeted fix and a heads-up to the crew is enough. Do not launch a program over a true one-off.

Not sure which it is? Log it in one line and watch. Two of the same near-miss is a pattern, and now you have the evidence.

Turn it into one specific change

A near-miss that ends in "everybody be more careful" changed nothing. Produce one concrete change with an owner and a by-when.

• Add the missed step to a checklist. Move a tool onto the standard truck load. Change the sequence so the hazard cannot be reached. Add a verification before sign-off.
• Right-size it. A one-off gets a verbal correction. A pattern gets a written change to the standard so it survives past memory and past this crew.
• Close the loop out loud. Tell the crew what changed and why. A near-miss that visibly fixed something is what teaches people to speak up next time instead of staying quiet.

The recap

1. Make the scene safe before anything else.
2. Capture the sequence and conditions while fresh.
3. Ask whether only luck saved it. If yes, treat it as a real failure.
4. Check whether it fits a pattern or is a true one-off.
5. Ship one specific change, sized to which it is, and tell the crew.

The judgment to bank: nothing broke is the reason to look harder, not the reason to move on. You will never get a cheaper warning than the one that cost you nothing.

References

• OSHA, Recommended Practices for Safety and Health Programs (hazard identification and near-miss investigation)
• See related: Near-Miss Reporting Without Blame; Treating a Near-Miss as a Free Warning
• See related: The Difference Between a One-Off and a Pattern; The Five Whys on a Service Call