Monitoring is frequently proposed as a safety improvement when the honest description is that it records the hazard more accurately.
The hierarchy, in order
Eliminate the hazard. Remove the task, automate it, redesign the flow.
Substitute something less hazardous.
Engineering controls: barriers, guards, separation, ventilation, mechanical assistance.
Administrative controls: procedures, training, scheduling, signage.
Personal protective equipment.
Monitoring sits between the last two, as a support to administrative control. It is close to the weakest thing available.
Where it gets substituted
Proximity detection instead of physical separation, which is the most common and most consequential substitution.
Speed monitoring instead of layout change, where the layout is what produces the speed.
Posture sensors instead of changing pick heights, which is measuring a problem that a slotting change would remove.
Alerting instead of ventilation.
Dashboards instead of staffing, where the hazard is fatigue from understaffing.
In each case the monitoring is cheaper and faster, which is why it wins, and it leaves the hazard in place.
The test
Would this control work if nobody looked at the output?
Barriers work when nobody is watching. Alerts do not.
A control that depends on a person reacting correctly, every time, under pressure, is an administrative control whatever technology delivers it.
Ask what the monitoring is warning about, and whether that thing could be removed. If it could, the removal is the control and the monitoring is an interim measure with an end date.
Alarm fatigue
Every warning system degrades with volume.
Operators stop reacting to alarms that fire constantly, which is a predictable human response rather than a discipline failure.
Measure the alarm rate and treat a high one as a defect in the system rather than in the people.
Differentiate severity, so an imminent collision is distinguishable from routine proximity.
A warning system with a high false rate is worse than none, because it trains people to ignore the class of signal.
Sensor defeat
Systems that impede work get defeated, by people who are otherwise conscientious.
Sensors covered, disabled, or routed around.
Which is a design finding: the system is imposing a cost the operation cannot absorb.
Detect it, and investigate why rather than treating it as misconduct, because the reason will be that the work cannot be done otherwise.
What monitoring genuinely contributes
Worth stating positively, because the list is real.
Finding conditions that produce near-misses, before an injury.
Quantifying the case for the physical control that should have been there.
Investigation evidence.
Verifying that a control is working, which is otherwise assumed.
Compliance evidence for pre-use checks, licensing and exposure limits.
Immediate warning where the hazard genuinely cannot be engineered out.
How to report it
Never as the control, always as a supplement with the primary control named.
With the physical work it is standing in for, and a date.
Otherwise the interim measure becomes permanent, which is the outcome the whole hierarchy exists to prevent.
Recording the interim measure
How to stop a supplement becoming the permanent control.
Name the primary control that should be in place: the barrier, the separation, the redesign.
Record the monitoring as interim, with a date for the primary work.
Put it on the risk register with that date, not as a closed item.
Review at the date, and if the primary work has not happened, record why and set a new date.
Report the count of interim measures past their date, which should be small and rarely is.
Investigating sensor defeat
Disabled sensors are a design finding, not a conduct finding.
Detect it: sensors covered, unplugged, or consistently absent from a route.
Ask why, without consequence, because the answer is what you need.
The reason is almost always that the system impedes the work — false alarms, a blocked path, a delay the operation cannot absorb.
Fix the impediment, then re-enable.
Treating it as misconduct produces better concealment, and the next defeat will not be detectable.