Near-Miss Signal Explained: 4 Evidence Levels That Separate Warning From Noise

F7 explainer for EHS managers, incident investigators, and frontline supervisors

A near-miss signal is evidence that an unwanted outcome was possible but did not occur, either because a barrier worked, the exposure ended early, or circumstances changed. The signal becomes decision-grade only when the investigation connects the event to a specific hazard, a tested control, the actual exposure, and the conditions that shaped the work.

A near miss is not automatically a warning, and a dramatic story is not automatically strong evidence. Across 25+ years of safety leadership, Andreza Araújo has consistently treated the quality of the evidence as more important than the label attached to the event. That distinction matters because a weak near-miss report can create activity without improving risk control.

Definition

A near-miss signal is a documented event in which a harmful outcome was possible, but the outcome was avoided. The avoided harm may result from a functioning barrier, a last-second recovery, a fortuitous change, or an exposure that stopped before contact. ISO 45001 specifies that organizations must identify hazards and assess risks, so the useful question is not only what happened, but which risk pathway was briefly exposed.

James Reason’s distinction between active failures and latent conditions helps keep the analysis broad. The person closest to the event may have made the final move, yet equipment design, unclear authorization, weak maintenance, or production pressure may have shaped the available choices. Andreza’s book Luck or Capability makes the same practical point in a different way, because an avoided outcome can reflect luck rather than reliable capability.

Level 1: The event is only reported

The first evidence level confirms that someone noticed an unusual condition, but it does not yet prove the hazard, exposure, or control status. A report such as “a tool almost fell” is a starting point, not an investigation finding. It needs location, task, object, height, people exposed, and the condition that prevented contact.

At this level, the supervisor should preserve the original account and ask clarifying questions without turning the conversation into a search for fault. HSE guidance on incident reporting emphasizes learning from events and dangerous occurrences, which means the report should remain specific enough for another person to test.

Level 2: The hazard and exposure are verified

The second evidence level shows what could have caused harm and who or what was exposed. Verification may come from photographs, equipment position, access records, task observations, or measurements that still represent the work condition. A near miss becomes more useful when the investigator can describe the credible injury or damage pathway in concrete terms.

For example, a suspended load that moved outside its planned path should be connected to the people within the possible line of fire, the lift plan, the exclusion zone, and the communication method. The report should state what was observed and what was inferred, because those are different forms of evidence.

Level 3: The barrier performance is known

The third evidence level identifies the control that prevented harm and shows whether it worked as designed. The control may be an interlock, guard, isolation, exclusion zone, permit condition, inspection, or supervisor intervention. A signed form alone does not prove performance, since documentation can show intent while the field condition shows what actually held.

Andreza Araújo’s safety-culture work places this distinction between compliance and control at the center of practical diagnosis. A site can complete the required record and still have a barrier that is unavailable, misunderstood, poorly maintained, or too easy to defeat. The investigation should therefore name the control, test its availability, and identify the evidence that supports the conclusion.

Level 4: The conditions behind the exposure are understood

The fourth evidence level explains why the exposure was possible in that work system. The investigation connects the event to design, planning, workload, competence, supervision, maintenance, contractor interfaces, or decision rights. ILO guidance on occupational safety and health supports this wider responsibility because prevention depends on organizing work safely, not merely correcting the final action.

This level does not remove accountability from the people involved. It prevents the report from stopping at “the procedure was not followed” when the procedure was unclear, the equipment did not match the task, or the operating plan made the safe route impractical. A strong finding states the condition, names the evidence, and assigns an owner who can verify that the exposure pathway has changed, which is the point at which a near-miss review becomes a control decision rather than a narrative exercise.

How to differentiate the four evidence levels

The four levels differ by what they allow a leader to decide. A report supports triage, verified exposure supports risk prioritization, known barrier performance supports control assurance, and understood work conditions support durable corrective action. The levels should not be confused because each one answers a different management question.

When a near miss deserves escalation

Escalate a near miss when it involves serious-injury or fatality potential, a failed critical control, repeated exposure, unclear ownership, or a condition that could affect more than one work group. The absence of injury should not lower the priority when the credible outcome was severe. Andreza’s work on safety culture returns to this principle: safety is about coming home, so the organization must examine the path that almost prevented that outcome.

The practical test is whether leaders can answer four questions with evidence. What hazard was present? Who or what was exposed? Which barrier held or failed? What condition made the exposure possible? If the report cannot answer those questions, it may still be worth recording, but it is not ready to guide a major corrective decision.