Most organisations take comfort from low incident rates. After all, this typically means fewer injuries, fewer lost days and fewer reportable events.
And yet, when a major incident occurs, the findings are rarely surprising. Investigations consistently show that the conditions were present long before the event: production pressure, warning signals ignored, degraded controls, unmanaged change etc. The incident was rare, but the precursors were not.
The Problem With Chasing Events
Major incidents are, by definition, infrequent. Using their absence as evidence of control is statistically weak.
Reviews such as the Baker Panel and multiple HSE learning reports have shown that organisations with strong personal safety records can still experience catastrophic failures. Frequency metrics hide exposure; they do not measure it.
One consistent finding across major accident investigations is that most serious incidents are preceded by multiple identifiable warning signs, often months or years in advance, and that were often documented but not acted upon.
Low incident rates often reflect luck, not resilience. Don’t shoot the messenger!
Risk Builds Quietly
Major accidents do not appear suddenly. They emerge from everyday conditions that become normal over time:
- Temporary controls becoming permanent
- Deviations becoming routine
- Assumptions replacing verification
These conditions are common, measurable, and often visible, but they are rarely tracked with the same discipline as injuries or near misses.
Why Organisations Are Still “Surprised”
The question asked after serious incidents is often, “How could this happen?”
The more accurate question is, “Why were we not paying attention to what was already there?”
HSE investigations repeatedly show that precursor signals existed but were not elevated, escalated, or acted upon. The system was functioning as designed, just not as intended.
A More Useful Measure of Safety
If organisations want to prevent major incidents, they must focus less on rare outcomes and more on common conditions:
- Workarounds becoming accepted
- Risk ownership becoming unclear (This is one I am particularly passionate about as many organisations have a culture of risk management being the responsibility of the EHS manager!!)
- Control degradation
- Unmanaged change
These are not hypothetical risks. They are present in most operations, most days.
The Real Test of EHS Leadership
Strong EHS performance is not the absence of incidents.
It is the ability to see risk building before something goes wrong. And right now that's not something that AI can be relied on for (but that's another day's topic).
At dulann, we believe the question is not whether a major incident will occur, but whether organisations are measuring and managing the conditions that make one possible.
Because major incidents are rare.
The conditions that create them are not.
