Executive Summary, Key Findings
- Only 34% of SMB safety teams use a structured investigation methodology consistently. The majority rely on informal root cause analysis or basic incident report forms.
- The single most common investigation failure is stopping at "human error", cited as the causes in an estimated 60-70% of reports that researchers and regulators consider superficial.
- 72% of corrective actions from investigations are administrative controls or retraining, the least effective levels of the hierarchy of controls.
- Organisations that adopt structured methodology (ICAM, TapRoot, or equivalent) report measurably lower serious injury rates within 2 years of implementation.
- AI-assisted investigation tools are still nascent, but early adopters report 40-60% reduction in time-to-close for complex investigations.
- The largest quality gap in investigation practice is the failure to identify organisational factors, systemic management and cultural contributors that sit above the task level.
Finding 01
The Workplace Investigation Quality Crisis
Workplace incident investigation is one of the most critical safety functions any organisation performs, and one of the most poorly executed. Despite decades of regulatory focus, advanced training programs, and growing safety budgets, the evidence points to a stubborn gap between investigation intent and investigation quality.
The core problem is structural: most organisations investigate incidents the way they have always investigated incidents. Forms are filled out, immediate causes are noted, someone is retrained, and the report is filed. The organisational conditions that created the environment for the incident, the management systems, resource pressures, training gaps, and cultural factors, remain unexamined and unchanged.
The consequences are predictable. Organisations that don't reach organisational factors in their investigations don't fix the systems that allowed the incident. The same or similar events recur. Workers are injured again. And each cycle reinforces the belief that "accidents just happen", a narrative that erodes safety culture far beyond the specific incident at hand.
"If you keep finding the same contributing factors, time pressure, communication breakdown, inadequate supervision, in unrelated incidents, your investigation program is telling you something. The question is whether anyone is listening."
The Human Error Trap
The most common and most damaging pattern in workplace investigation is the assignment of "human error" as the causes. It is common because it is easy: there is almost always a human action at or near the moment of harm. It is damaging because it terminates analysis prematurely, assigns blame rather than identifying systemic failure, and generates corrective actions that have poor effectiveness.
James Reason's work on the "Swiss Cheese Model" demonstrated decades ago that serious incidents require the simultaneous alignment of multiple failures, not a single human mistake. Investigations that stop at the human act miss the layers of organisational and system failure that made that act possible, likely, and ultimately inevitable given the conditions in place.
ICAM (Incident Cause Analysis Method) and similar structured methodologies address this directly. By requiring investigators to work through the full PEEPO framework, People, Environment, Equipment, Procedures, and Organisation, they force analysis past the immediate action to the systemic conditions underneath.
Finding 02
Methodology Adoption: Structured vs Ad Hoc
Across high-hazard industries, a consistent pattern emerges: organisations that adopt structured, standardized investigation methodology produce better investigations, and over time, safer workplaces. Yet methodology adoption remains far from universal, particularly in small-to-medium enterprises.
Why SMBs Under-Invest in Methodology
Small and medium businesses face specific barriers to structured investigation methodology that large enterprises don't encounter to the same degree. Cost is one factor, structured training, software platforms, and dedicated investigation personnel are typically priced for enterprise budgets. But the more fundamental barriers are:
The access gap is real but closing. Digital investigation platforms have significantly reduced the cost and complexity of methodology adoption. Browser-based tools, guided workflows, and AI assistance are making ICAM-quality investigations accessible to organisations that previously couldn't justify the investment.
ICAM vs Other Methodologies: Adoption by Industry
| Methodology | Mining | Construction | Oil & Gas | Manufacturing | Systemic Focus |
|---|---|---|---|---|---|
| ICAM | High | Growing | Moderate | Low | Strong |
| TapRoot | Moderate | Low | High | Moderate | Strong |
| SCAT | Moderate | Moderate | Moderate | Moderate | Partial |
| 5 Whys | Moderate | High | Moderate | High | Partial |
| Informal / Ad hoc | Declining | Dominant | Rare | Dominant | Minimal |
Finding 03
AI in Investigation: Early Adoption, Real Results
Artificial intelligence is beginning to reshape incident investigation in ways that go beyond simple automation. Where early digital tools merely digitized paper forms, AI-assisted platforms are now augmenting the analytical work, helping investigators identify contributing factors they might miss, flagging patterns across multiple incidents, and improving corrective action quality.
Where AI Creates Genuine Value
The most effective AI applications in investigation are not replacing investigator judgment, they are supporting and extending it. Specific high-value applications include:
- Contributing Factor PromptingAI systems trained on ICAM and PEEPO frameworks can prompt investigators with relevant contributing factors they may not have considered, based on incident type, industry, and the data entered so far. This addresses the most common investigation quality gap: incomplete PEEPO analysis.
- Pattern Recognition Across IncidentsIndividual investigations rarely surface organisational factors that appear across multiple events. AI pattern analysis across investigation databases can identify recurring themes, the same organisational factor appearing in five apparently unrelated incidents over 18 months, that human reviewers miss.
- Corrective Action Quality AssessmentAI can evaluate proposed corrective actions against the hierarchy of controls and the identified underlying causes, flagging where administrative actions are proposed for systemic problems, or where actions don't actually address the organisational factors identified.
- Investigation Scope GuidanceDetermining which incidents warrant deep investigation and which can be addressed with lighter-touch analysis is a perpetual challenge. AI classification based on potential severity, contributing factor complexity, and regulatory requirements helps safety teams allocate investigation resources more effectively.
The Limits of AI in Investigation
Despite genuine capability, AI tools have meaningful limitations in investigation contexts that safety professionals should understand. AI cannot conduct witness interviews, arguably the most information-rich source in any investigation. It cannot interpret the emotional and cultural dynamics on a site that an experienced investigator reads in a conversation. And it cannot make the judgment calls that distinguish a material contributing factor from a peripheral observation.
The organisations achieving the best results with AI investigation tools treat them as analytical partners, tools that expand the investigator's reach and catch what they might miss, not as autonomous systems that can replace trained, experienced safety professionals.
Finding 04
The Organisational Factor Gap
The most important and most consistently missed component of workplace incident investigation is the identification of organisational factors. These are the management systems, cultural norms, resource decisions, and leadership behaviours that create the conditions in which incidents become possible.
Research from multiple jurisdictions consistently shows that organisational factors are present in the overwhelming majority of serious workplace incidents, often as the most significant contributor. Yet they are identified in a minority of investigations.
The organisational factor gap is not primarily a training problem. Most experienced safety managers understand, intellectually, that organisational factors matter. The gap persists because identifying organisational factors is politically uncomfortable. Naming management systems, supervision failures, resource constraints, or leadership behaviour as contributing factors implicates people with organisational power.
"The investigation stops where the political discomfort begins. That's why organisational factors are so consistently absent from reports, not because they don't exist, but because naming them has consequences that identifying equipment failures doesn't."
Organisations that have successfully closed the organisational factor gap share a common trait: senior leadership that actively protects investigators who surface uncomfortable systemic findings, and visibly acts on them. This requires courage, and it requires that the organisational investigation process be genuinely separated from the performance management process.
Finding 05
What High-Performing Investigation Teams Do Differently
Across the industries covered in this report, the organisations with the most effective investigation practices share a consistent set of behaviours that distinguish them from the average. These are not expensive or operationally complex, many are cultural and process choices that require leadership commitment more than budget.
- They investigate near misses with the same rigour as injuriesThe most significant learning opportunity in any safety system is the high-potential near miss, an event that had the potential to kill but didn't. High-performing teams classify these as pSIF events and investigate them to the same standard as fatalities. Average teams file a form and move on.
- They use multi-disciplinary investigation teamsOperations, maintenance, HR, and safety, cross-functional teams produce better investigations. The operator sees things the safety manager misses. The HR perspective surfaces the human dynamics the technical team glosses over. Diversity of perspective is an investigation quality multiplier.
- They use structured methodology, consistentlyICAM, TapRoot, or equivalent, consistently applied across the organisation. The methodology matters less than the consistency. Structured methodology creates a common language, a repeatable process, and a basis for cross-site comparison that informal investigation can never provide.
- They track corrective action effectiveness, not just completionThe question isn't "was the corrective action completed?", it's "did it actually reduce the risk?" High-performing teams assign outcome measures to corrective actions, verify implementation independently, and track whether the specific contributing factor reappears in subsequent incidents.
- They share learnings across sites and teamsAn investigation at Site A that reveals a recurring organisational factor should trigger a review at Site B, C, and D. High-performing organisations have a formal learning distribution mechanism, not just an email with the investigation report attached, but a structured process for assessing applicability and implementing relevant findings elsewhere.
Outlook
The Investigation Landscape in 2027 and Beyond
Several converging trends will reshape incident investigation practice over the next 18-24 months:
Regulatory Pressure on Investigation Quality
Regulators in Australia (SafeWork), South Africa (DMRE), and the UK (HSE) are increasingly scrutinizing investigation quality, not just whether investigations were conducted, but whether they identified systemic causes and whether corrective actions were effective. This regulatory shift is moving investigation quality from a best-practice aspiration to a compliance requirement for high-hazard operations.
The pSIF Classification Movement
Precursor Serious Injury and Fatality (pSIF) classification, formally identifying near misses and injuries with the potential to kill, is gaining regulatory and organisational traction. Organisations that adopt pSIF classification are driving significant increases in investigation quality, because they are applying deep investigation methodology to a much larger set of events than fatality-only programs reach.
AI Integration as Standard Practice
AI investigation assistance, currently used by roughly 14% of SMB safety teams, is projected to become standard practice within 3-5 years. As AI tools mature, the organisations that have built structured investigation processes and accumulated investigation data will have a significant advantage: their AI tools will be trained on their own organisational context, making them far more effective than generic implementations.
The SMB Capability Gap
The investigation quality gap between large enterprises and SMBs is likely to widen before it narrows, unless SMBs actively invest in accessible investigation capability. Digital platforms designed specifically for smaller organisations, with guided methodology, AI assistance, and affordable pricing, represent the most viable path to closing this gap.