A near miss is an event that could have caused harm, loss or disruption but did not on that occasion. For a security team, that might be an unsecured access point discovered before entry occurred, a vehicle conflict without impact or a failed control caught during patrol.
Near-miss reporting is useful when it produces learning rather than blame. The process should be easy enough to capture weak signals and structured enough to route material risks for action.
Define what belongs in the process
Give staff concrete examples relevant to their sites and contracts. Distinguish a near miss from routine maintenance, a security observation and an incident, but allow a reviewer to reclassify the report without punishing the person who raised it.
The reporting threshold should favour useful visibility. If staff fear discipline for a good-faith report, the organisation will lose the signals it needs most.
Capture conditions and controls
- What could have happened
- What condition or action created the exposure
- Which existing control failed or was absent
- What prevented harm this time
- What immediate action made the area safe
- Who owns longer-term follow-up
Avoid stopping at individual error
A useful review asks why the condition was possible: Was the procedure clear? Was the equipment maintained? Did workload, access or supervision contribute? The objective is to find controllable factors and reduce recurrence, not to make the report sound conclusive.
OSHA's incident-investigation guidance similarly advises looking beyond the immediate trigger to underlying program shortcomings and root causes.
Close the feedback loop
People stop reporting when nothing visibly happens. Acknowledge the submission, assign the follow-up, record the decision and tell affected teams what changed. Not every near miss needs a large investigation, but every report should have a clear disposition.
Review patterns without gaming the metric
A rising near-miss count may show greater exposure, better reporting or both. Do not treat fewer reports as an automatic success. Review category, site, control failure, recurrence and time to action alongside the raw count.