Where Near Miss Counts Can Mislead Assurance

Near miss data can look mature in an audit pack while still saying very little about why the work created the event path. The count is often less useful than the variation behind it.

This is not an argument against near miss reporting. Auditors understand evidence. What operations can add is a feel for where the evidence can become thin, especially when the report looks complete but the work behind it is still unclear.

A near miss register can be full and still weak. It can show activity without showing learning. It can show categorisation without showing whether the organisation understands where work is departing from its intended method. That is the discomfort worth paying attention to: the report may describe the almost-event, while the system still misses the reason the event path opened.

Where the count starts to feel thin

Near miss volume is often presented as a sign of reporting culture. Sometimes that is fair. But from inside the work, the same number can mean different things in different places. High reporting may reflect trust, or it may reflect a campaign. Low reporting may reflect good control, or it may reflect silence.

The more useful line of enquiry is not whether the number is high or low. It is whether reported events are being converted into understanding about work conditions, control reliability, and the standards people are actually using.

That is where a sample of reports can say more than the trend. If the report says a hammer fell and no one was injured, the near miss category has done its basic job. The next question is different: why was the hammer able to fall? Was the tool secured? Was the work method followed? Was the standard workable for the task? Was the condition created by the plan, the setup, the supervision, or the pressure of the job?

Why deviation is a different measure

A near miss is outcome-adjacent. It is usually defined by the injury, damage, or loss that did not occur. That makes it useful as a prompt, but limited as a measure. It tells you something nearly happened; it does not necessarily tell you why the pathway existed.

Deviation from Standards is a different class of measure. It does not depend on whether an injury almost occurred. It asks whether work departed from the intended method, and what that departure reveals. Was reliability decreased? Was vulnerability increased? Did the system absorb variation? Did the crew find a better way of working?

That distinction matters. Near miss looks at the event boundary. DFS looks at the work pattern. One may record that the hammer fell without injury. The other asks what variation in the work allowed the hammer to fall in the first place.

Where systems often fail

A near miss system can look controlled because everything is closed. That is not the same as learning. Closure often means the record has been assigned, actioned, and completed. It does not necessarily mean the organisation understood why the deviation occurred.

Repeat patterns, repeated locations, repeated standards, and repeated corrective actions are often where the operational discomfort sits. If the same type of report keeps appearing, the issue is probably not the individual event. It may be a condition in the work system that has not been changed.

A useful way to feel that difference is to follow a sample of reports back to the workface. Ask what changed before the event became possible. Ask how the task is normally done. Look at whether the standard fits the task. Then compare that reality with what the report and action close-out say.

What assurance can draw from it

The aim is not to make near miss reporting heavier. In many organisations it is already too heavy. The aim is to make it more useful.

Assurance can help by giving more weight to the stronger signal: clearer links to the intended standard, better classification of the variation, proportional response, and visible learning from repeated deviations.

Near miss reporting still has value. The nudge from operations is simply this: do not stop at the almost-event. Use it as an entry point, then look for the deviation that explains why the almost-event was possible.