Airlines Have More Safety Data Than Ever. Why Aren’t They Learning Faster?

6 minute read

Most airlines operating today have more safety data than at any point in their history. Serious event reports, near miss records, audit findings, corrective actions, and investigation outputs. Years of it, accumulated across fleets, bases, and operational teams. And somewhere inside that data, the conditions that will contribute to your next serious event are already present.

The question is whether you are set up to find years data that simply cannot be interrogated before they combine into something significant.

At COMET, we work with organizations across the world’s most demanding high-risk industries.

The pattern we see is consistent regardless of sector: more reporting does not automatically mean more learning.

For airlines, that gap between investigation activity and meaningful safety improvement creates the potential for the next serious injury or fatality, or serious operational failure. It is why our work with Intelex and Arcadis on a combined airline SMS offering matters. And it starts with three structural challenges that no amount of reporting volume will fix on its own.

The data going in is not good enough

An investigation can only ever be as good as the evidence it is built on. That sounds obvious. In practice, it is one of the most consistently underestimated challenges in safety management, and I say that having worked with some of the most safety-conscious organizations on the planet.

Evidence degrades fast, witness recall fades within hours, physical conditions change, and operational records get updated. The window for capturing high-quality, objective, multi-source data is narrow, and when it closes, investigations default to what is easiest to document rather than what is most important to understand.

Data quality is not just about speed; it is about structure.

One of the most damaging patterns we see across safety programs is years of data that simply cannot be interrogated at scale, because it was captured to be stored, not to be further analyzed.

Free text descriptions. Inconsistent classification, terminology that varies by base, by team, by individual investigator. The data exists. The intelligence locked inside it does not. And you cannot act on intelligence you cannot understand.

Before any organization asks how to reduce serious events, they need to ask a harder question first:

• If you needed to identify the causal factors appearing most frequently across your last five years of events, how long would that take? And how confident would you be in the answer? For most, it is not a comfortable exercise.

This matters just as much for a fifty-aircraft regional carrier as it does for a global network airline. The volume differs, but the problem does not.

Human error is where investigations stop. It should be where they start.

Ask most aviation safety professionals where investigations tend to fall short, and they will give you the same answer: the treatment of human error as a conclusion rather than a starting point.

A crew member deviated from procedure, a ground handler missed a step, a technician made a judgment call under pressure. These are findings. They are not root causes. Root causes come later, but are definitely influenced by a deeper understanding of the reasons why humans are fallible. And until an investigation gets there, the conditions that produced the failure stay firmly in place. The next event is already in motion.

Repeat events are a system problem

I have lost count of the investigations I have seen where the report was closed, the actions were assigned, and within months a near-identical event occurred involving different people, different locations, and the same underlying causes. That is not bad luck. That is what happens when investigations stop at the individual rather than the system.

Structured human factors analysis, embedded directly into the investigation workflow, is what changes this. It is not an add-on or a workstream reserved for the most serious accidents. It is the methodology that separates investigations that reduce risk from investigations that merely document it.

For smaller carriers without dedicated investigation teams, this point carries particular weight. The quality of an investigation should not depend on the experience level of whoever happens to be available to lead it. A structured framework levels that. Every investigation, regardless of severity or the seniority of the investigator, should produce findings that are rigorous, consistent, and comparable across the whole operation. Anything less is a missed opportunity.

Your safety data is not telling you what it knows

The third gap is arguably the most consequential, and the one where most safety programs are furthest behind the opportunity now available to them.

Organizations that have genuinely invested in strong reporting and investigation processes are sitting on years of structured safety data. The patterns inside that data, which causal factors recur across high-potential events, where systemic risk is accumulating, which interventions are actually working are almost entirely invisible to the people who need them most. Not because the data is not there, but because the tools to surface it have not been.

That is changing. The COMET AI Assistant is purpose-built for investigation and root cause analysis. It works across an organization’s full investigation data set, answering questions that have previously had no practical answer: where is risk building? What is driving repeat failures? What does strong performance look like in our best operational areas, and what would it take to replicate it elsewhere? All through natural conversation. All without needing a data analyst in the room. Goodbye PowerBi.

For a major carrier managing thousands of events annually across a global network, that capability is transformative. For a regional operator with a lean safety team and a finite prevention resource, it is arguably even more valuable. The question I would ask any safety leader is simply this: is your safety data currently working as hard as your safety team? Because in most organizations, the honest answer is no.

Related article: From Incident Management to Incident Prevention

Closing the loop

Every investigation represents a commitment. A commitment to understand what happened, why the system allowed it, and what needs to change. When that commitment is supported by quality data, structured methodology, and the intelligence to see across the full body of evidence, investigations stop being a record-keeping exercise and start being the most powerful prevention tool an organization has. That is the standard the combined Intelex, Arcadis, and COMET offering is built to deliver.

Intelex captures and manages safety data at enterprise scale. COMET brings the investigation methodology, human factors analysis, and an AI-assisted intelligence layer. Arcadis ensures the technology is configured, embedded, and adopted across operational teams, not simply deployed and left to find its own level.

The result links every report to a root cause, every root cause to a preventive action, and every preventive action to a measurable outcome.

Aviation has always held itself to a higher standard than most industries. The investigation capability now available means there is no longer any reason for that standard to stop at the point where the real work begins. The only gap worth closing now is the one between data and insight.

To find out more about the combined Intelex, Arcadis, and COMET airline SMS offering, download the one-pager here.