Most organisations have rules that tell them when to investigate an incident. The more serious the outcome, the more significant the investigation. Add potential consequences into the mix, and the investigation can quickly become larger again. It sounds logical. The problem is that the severity of an outcome does not tell us how much we can learn from an event.
Consider two incidents that both result in a broken arm. In one, someone trips while walking through an empty area. In the other, a worker is climbing into a truck when a handrail breaks. The injury is the same. The learning opportunity is not.
The second event might raise questions about equipment condition, maintenance, access, work design or whether the controls we thought were in place were actually effective. The first may tell us very little that we do not already know.
Research has questioned the value of measures such as TRIR and TRIFR as indicators of safety performance. Dekker and Tooma (2021) point out that these methods provide limited insight into the hazards and risks behind the number. Hallowell et al. (2020) similarly found no discernible association between TRIR and fatalities.
So, if injury severity is not enough, what should determine the level of investigation?
A better question is: What are we actually trying to learn?
Some events clearly warrant a high-level investigation. A fatality, serious injury or event with credible potential for a critical outcome should attract significant attention. Events involving unusual conditions, poorly understood interactions, or a challenge to how a critical risk is being managed may also justify going deep.
At this level, we need to understand how the organisation planned the work, how it was actually performed, what conditions shaped decisions, how controls interacted and whether similar vulnerabilities exist elsewhere. But not every event sits at that level.
An event involving the failure or absence of a particular control might only require a focused inquiry. We may need to understand why the control was unavailable, why it did not work as expected, or whether the assumptions behind it remain valid. That might require a formal investigation. Organisations might better address control failure through a structured discussion with those who understand the work.
For other events, particularly minor and medium-level ones, a shorter learning conversation may be enough. We want to understand what was different on the day, what people had to adapt to, or whether there is a practical improvement worth making. That is still learning. It doesn’t need to become a 30-page investigation report.
And then there are events where we should seriously consider not investigating at all. Sometimes an event is minor, the circumstances are well understood, the controls are known, there is no credible risk of critical failure, and there is little reason to believe another investigation will teach us anything new.
In those circumstances, record the event and move on. Recording still matters. Individual events that tell us very little can become meaningful when looked at together. If we start seeing the same event, involving the same equipment or under similar conditions, the pattern may tell us something the individual event did not. The learning opportunity has changed.
Rather than having only two choices, investigate or don’t investigate, organisations should think in terms of different levels of inquiry: a high-level investigation where there is significant risk and uncertainty; a focused investigation into a particular control or issue; a learning conversation or group reflection where operational insight is needed; or simply recording and monitoring the event. The difficult part is deciding where the line sits.
That decision should not rest only on an injury classification or matrix. It should consider credible potential, the importance of the controls involved, whether something genuinely unexpected occurred, whether the learning could apply elsewhere, and whether the organisation is likely to know more after the inquiry than it knew before.
Who makes that decision matters too. Safety may see one thing. Operations may see another. The people doing the work may see something that both have missed. None of this removes legal or regulatory requirements. Where an investigation must be conducted, it needs to happen.
Investigation resources are not unlimited. Every hour spent investigating something we already understand is an hour we can’t spend understanding something we don’t. The objective should not be to investigate more incidents. Nor should it be to investigate fewer. It should apply the right level of inquiry to the right event. Go deep when there is something important to understand. Use a lighter approach when that is enough. And when there is genuinely nothing new to learn, have the confidence not to investigate at all.
Because completing an investigation is not the measure of success, learning something that helps us manage future risk is.
Sources
Dekker, S. W. A., & Tooma, M. (2021). A capacity index to replace flawed incident-based metrics for worker safety. International Labour Review. https://doi.org/10.1111/ilr.12210
Hallowell, M., Quashne, M., Salas, R., Jones, M., MacLean, B., & Quinn, E. (2020). The Statistical Invalidity of TRIR as a Measure of Safety Performance. Construction Safety Research Alliance.
Alston, M., & Havinga, J. (2024). Event Learning Assessment: Being smart about what to investigate. Investigations Differently.
Contributor details:
Mark Alston | Director, Investigations Differently
Contributor bio: Mark Alston is Director of Investigations Differently and has more than two decades of experience in workplace investigations and risk management. He works with high-risk organisations across Australia and internationally, centring on practical investigation methods, organisational learning, and improving how organisations understand and manage risk.
LinkedIn: https://www.linkedin.com/in/alstonmark
