The Limits of Lagging Indicators: Why Traditional Safety Metrics Can’t See the Risk That Matters
August 19, 2026
8 minute read
Somewhere in a boardroom this quarter, a safety leader is presenting a slide that says the organization is winning. Recordable rates are down. DART is trending in the right direction. The chart is green, the arrow points where arrows are supposed to point, and everyone around the table nods and moves to the next agenda item.
Somewhere in that same organization, conditions for a serious injury or fatality are quietly assembling during a night shift, a maintenance turnaround, or a routine task that has been done a thousand times. None of it shows up on that slide.
The question that kept coming up at EHS Congress
This was the major theme we heard at this year’s EHS Congress in Berlin. Speaker after speaker, panel after panel, the conversation kept circling back to the same uncomfortable question: Have we built reporting structures that tell boards, investors, and shareholders what they want to hear, while losing visibility into the operational conditions that actually create risk?
I want to be careful here, because this is where the conversation usually goes off the rails. Nobody in Berlin was accusing the industry of cooking the books. The debate wasn’t about manipulation. It was about something more subtle and, frankly, more dangerous: incentive structures that reward organizations for measuring what’s easy to measure rather than what genuinely reflects operational resilience and risk exposure.
That’s not fraud. That’s drift.
And drift is harder to fix because everyone involved believes they’re doing the right thing.
The comfort of the countable
I’ve spent nearly four decades leading safety and health, most of it inside Fortune 500 operations, and I understand the gravitational pull of the lagging indicator better than most. Incident rates are countable. They’re comparable across sites and business units. They roll up neatly. They fit on one slide. When the CEO asks how safety is doing, a number gives you an answer in five seconds.
But here’s what that number actually tells you: how many injuries met a recordkeeping threshold during a trailing period. That’s it.
It tells you almost nothing about whether the next serious event is close or far away because SIF research has been clear for years. The precursors to life-altering events live in a different population than the slips, strains, and lacerations that dominate most incident logs. You can drive your recordable rate toward zero and leave your fatality exposure completely untouched. Many organizations have learned this lesson the hard way.
The deeper problem is what a low rate does to organizational attention. A green dashboard doesn’t just fail to reveal risk, it actively signals that nobody needs to go looking for it. The metric becomes a sedative. The leadership team relaxes. Resources shift elsewhere.
And the gap between work-as-imagined and work-as-performed keeps widening in the dark, because no one is measuring the thing that matters: what it actually takes to get the job done on a Tuesday when the parts are late, the procedure doesn’t fit the field condition, and two people are covering the work of three.
A dashboard can’t walk the floor. It never could.
What Berlin got right
The strongest sessions in Berlin weren’t the ones tearing down TRIR (that critique is a decade old and, at this point, it’s the easy part). The strongest sessions were the ones wrestling with what comes next: human factors, learning teams, SIF precursor identification, and the discipline of organizational learning as an operational capability rather than a post-incident ritual.
The thread connecting all of it was a shift in the fundamental question we ask:
| Traditional metrics | Emerging discipline | |
|---|---|---|
| The question asked | “What happened?” | “What are the conditions under which work succeeds, and where are those conditions eroding?” |
| What you measure | Reported events | Normal work |
| What you investigate | Failures | Successful-but-difficult tasks |
| What a quiet month means | Good news | An open question: Genuine control, or the absence of bad luck? |
Traditional metrics ask, “What happened?” The emerging discipline asks, “What are the conditions under which work succeeds, and where are those conditions eroding?”
That second question changes everything about what you measure:
- Instead of counting reported events, you start examining normal work.
- Instead of investigating only failures, you convene learning teams around successful-but-difficult tasks and ask the people who do them where the friction lives.
- Instead of treating a quiet month as good news, you ask whether it reflects genuine control or just the absence of bad luck. Those two things look identical on a lagging chart and could not be more different in reality.
I’ve sat in learning team sessions where a crew described, calmly and without a trace of complaint, a routine workaround that would have shocked me if I’d seen it captured in an incident report.
No event had occurred. Nothing was reportable. By every traditional measure, that task was a success story. But the operational condition — the gap between the written procedure and the real work — was a SIF precursor that was not visible, and no lagging indicator on earth was ever going to surface it. The only instrument sensitive enough to detect it was a conversation.
That’s the insight the Congress kept returning to: the future of EHS may be less about reporting incidents and more about understanding the operational conditions that allow organizations to learn, adapt, and prevent serious events before they occur. Not instead of measurement, but through better measurement.
The board conversation we need to have
Here’s where I’ll push past where some of the Berlin discussion stopped, because I think senior practitioners have an obligation the conference circuit sometimes lets us dodge: this problem is ours to fix, and the fix runs through the boardroom, not around it.
Board aren’t the villains. We trained them.
It’s tempting to frame boards and investors as the villains. They’re the audience demanding simple numbers that force us into shallow reporting. My experience says otherwise. Boards ask for incident rates because that’s what we’ve trained them to expect. For decades, we handed them a single lagging number and told them it meant safety. They believed us. The reporting structure that Berlin critiqued is a structure our profession built.
Which means we can rebuild it.
What risk-based safety reporting looks like
Governance audiences are perfectly capable of digesting risk-based reporting, they do it every quarter with financial exposure, market risk, and cyber threats. What they need from us is a coherent alternative. A balanced view that pairs outcome data with the leading signals that actually predict resilience:
- SIF precursor identification and closure rates.
- Verification of critical controls on the tasks that can kill someone.
- Learning team frequency and, more importantly, what the organization changed as a result.
- The health of the reporting culture itself. When hazard reporting drops but injury rates hold steady, don’t read that as stability. It usually means people have stopped telling you things.
The organizations doing this job well aren’t drowning their boards in data. They’re changing the narrative from “here’s what happened to us” to “here’s how well we understand and control our most serious exposures.” One is a scoreboard. The other is a risk conversation. Directors, in my experience, are hungry for the second one, they’ve just rarely been offered it.
Measuring what matters
I don’t want to leave the impression that lagging indicators should be thrown out. They shouldn’t.
Outcomes matter, and pretending otherwise is its form of drift. The failure isn’t in measuring results, it’s in measuring only results and mistaking that partial picture for the whole truth about operational risk.
The path forward is balance:
- Lagging indicators for accountability.
- Leading indicators for prediction.
- Learning mechanisms for the risk that no metric will ever capture on its own.
That balance doesn’t happen by accident. It has to be designed, defended, and reported with the same rigor we’ve historically reserved for the injury rate.
The message in Berlin was clear, even if it was uncomfortable: the metrics that made our profession credible are no longer sufficient to make our organizations safe. The most important risk has always existed below the reporting threshold — in normal work conditions, in the gap between procedure and field, and in the things people adapt to every single day without a word.
Our job now is to build measurement systems that are honest enough to see it.
If the themes from the 2026 EHS Congress convinced you the recordable rate isn’t enough, the next question is what to do with everything you’re already measuring. Our report, Turning EHS Data Collection Into Meaningful Action, shows safety leaders how to move beyond the metrics, convert data into meaningful narratives, and build a culture of proactive risk management.
