Underreporting in Safety: 7 Signals Your Metrics Are Too Clean

A safety dashboard can look excellent for the wrong reason. The recordable rate falls, near misses disappear, first-aid cases become rare, and leadership begins to believe the operation is finally stable.

The field may be telling a different story. Supervisors still hear about pain after manual handling, contractors still correct permits at the last minute, operators still work around failing equipment, and maintenance still knows which guard, valve, or sensor is becoming unreliable. The difference is that fewer weak signals are reaching the formal system.

This article is written for EHS managers, plant managers, operations directors, and senior leaders who need to test whether clean safety metrics represent real control or quiet underreporting. The thesis is direct: a clean dashboard is only credible when reporting volume, field evidence, and control verification move in the same direction.

Why underreporting is a leadership metric

Underreporting in safety is not merely an employee behavior problem. It is a measure of what the organization rewards, tolerates, complicates, and punishes after bad news appears.

If reporting creates an investigation that feels like blame, a supervisor lecture, a lost bonus, or a production delay without visible correction, people learn to manage safety information informally. They still talk about risk, but they talk in the break room, in the maintenance shop, or inside the crew. The dashboard receives only the part of reality that feels safe to disclose.

As Andreza Araujo argues in Safety Culture: From Theory to Practice, safety culture is visible in the gap between declared values and repeated behavior. A company can say every incident must be reported while its local reactions teach workers that reports create trouble for the person who speaks first.

That is why underreporting belongs in the executive safety conversation. It tells leaders whether the organization can hear weak signals while they are still cheap enough to correct.

1. Near misses fall while field exposure stays the same

The first signal is a drop in near-miss reporting without a believable change in exposure, work volume, contractor activity, maintenance backlog, or high-risk tasks. When the operation is just as busy but the weak-signal flow dries up, leaders should ask what changed in the reporting climate.

Near misses are not perfect predictors, but they are useful because they reveal what workers choose to tell the system before harm occurs. Heinrich and Bird's pyramid models are often debated in detail, yet the practical lesson remains valuable: precursor events provide management with chances to learn before serious loss.

The article on near-miss reporting myths explains why volume alone is not enough. A healthy system needs enough reports to expose weak signals and enough quality to show which controls are failing. A sudden reduction in both volume and detail should not be celebrated quickly.

The trap is calling silence maturity. Mature safety systems do not become quiet because people stop seeing risk. They become more precise because people report useful signals and leaders close the loop.

2. Recordable rates improve after incentive pressure

The second signal appears when TRIR, DART, or lost-time indicators improve soon after a bonus cycle, leadership campaign, audit period, customer visit, or public target. The improvement may be real, although the timing deserves scrutiny.

When leaders attach status, money, or reputation to a clean number, they may unintentionally make reporting feel disloyal. A worker with shoulder pain waits until the weekend. A supervisor encourages first aid instead of medical evaluation. A contractor does not mention a minor fall because the project is close to a milestone.

This connects with DART rate distortions and OSHA 300 log signals. Recordkeeping definitions matter, but culture decides whether the case reaches the definition honestly. A technically compliant log can still hide a system that discourages disclosure.

In her Portuguese title Muito Alem do Zero, or Far Beyond Zero, Andreza Araujo critiques the way zero-accident targets can distort behavior when leaders worship the number more than the learning. The problem is not ambition. The problem is making a clean metric more valuable than a truthful one.

3. First-aid cases and discomfort reports do not match the work

The third signal is a mismatch between the work performed and the low number of first-aid cases, discomfort reports, ergonomic complaints, or health-room contacts. Manual handling, repetitive work, heat, awkward postures, noise, chemicals, and long shifts normally create some flow of weak health signals.

If those signals are absent in a demanding operation, the reporting pathway may be too formal, too slow, too exposed, or too closely tied to discipline. Workers may self-treat, swap tasks quietly, borrow medication, or ask a colleague for help instead of entering the official system.

This matters because occupational health indicators often precede injury statistics. A site that does not hear discomfort early will eventually hear about restricted work, lost time, turnover, fatigue, or quality failures. The delay makes prevention harder and more expensive.

Across 25+ years in executive EHS roles, Andreza Araujo has seen that weak reporting is often defended as discipline. Managers say people should not exaggerate. The better question is whether the organization has made early reporting useful enough that workers do not wait until pain becomes incapacity.

4. Stop-work authority exists but is rarely used

The fourth signal is a stop-work policy with almost no stop-work events. In a complex operation, occasional pauses are normal evidence that workers and supervisors are seeing changing conditions and acting before loss.

A site with high-risk work, contractors, maintenance, energized systems, lifting, confined spaces, hot work, or vehicle interaction should not expect every job to flow without interruption. If nobody stops work, leaders should ask whether conditions are truly stable or whether stopping work has become socially expensive.

The article on stop-work authority makes the key point: authority is proven after someone uses it. If the worker who pauses a job receives sarcasm, isolation, schedule pressure, or a reputation for being difficult, the next person will solve the problem quietly or keep working.

Zero stop-work events in high-risk work can be a warning sign, not a badge of maturity. Leaders should expect enough pause-and-fix behavior to prove that the authority is alive.

5. Supervisor notes are richer than the official system

The fifth signal appears when supervisor notebooks, shift handovers, WhatsApp groups, maintenance requests, and informal crew conversations contain better risk information than the official safety database.

This split often emerges because the formal system is slow or punitive while informal channels are fast and practical. A supervisor may know that a task is drifting, a contractor is struggling, or a control is weak, but the organization receives only a cleaned-up version because the reporting tool asks for categories that do not fit real work.

James Reason's work on latent failures helps explain the danger. Many serious events are prepared by conditions that are known locally before they become visible corporately. If the official system cannot absorb local knowledge, leaders lose the chance to correct the latent condition.

A strong EHS manager should compare formal records with informal field evidence. The goal is not to police private notes. The goal is to learn why the official system is less useful than the workaround.

6. SIF precursors are missing from routine reports

The sixth signal is the absence of serious injury and fatality precursors in routine reporting. A site may report slips, minor cuts, and housekeeping observations while missing dropped objects, failed isolations, bypassed guards, uncontrolled energy, line-of-fire exposure, unstable loads, and permit failures.

That pattern creates a dangerous illusion. The reporting system looks active because many low-severity items are entered, but the signals most connected to fatal risk remain invisible. Leaders then allocate attention to the easiest reports rather than the most consequential controls.

This is why SIF precursor metrics should sit beside traditional rates. They help leaders ask whether high-consequence exposures are being seen, reported, classified, and corrected before the organization learns through serious harm.

In more than 250 cultural transformation projects supported by Andreza Araujo's work, a repeated weakness is the confusion between activity and insight. Many reports do not prove learning if the organization is not capturing the signals that can kill.

7. Corrective actions close without report volume changing

The seventh signal is corrective-action closure that does not change reporting behavior. If leaders repair a condition, remove a barrier to reporting, simplify a form, coach a supervisor, or protect workers after bad news, the reporting pattern should eventually show it.

When actions close but reporting remains thin, the organization may be closing tasks rather than changing trust. The database says done, while the field still believes that speaking up creates personal cost and little improvement.

Use corrective-action closure metrics to test whether risk changed after the action. Add one question that many dashboards omit: did the exposed group start reporting earlier, better, or more specifically after the fix?

Andreza Araujo's Safety Culture Diagnosis approach is useful here because it does not treat culture as a slogan. It asks whether the practiced routine changed. In underreporting, the practiced routine is visible in what people choose to disclose after leadership has had several chances to prove it can hear the truth.

Underreporting signals compared

Underreporting rarely announces itself as concealment. It appears as inconsistencies between the formal record and the work system around it.

How to audit reporting honesty in 30 days

Start with one high-risk area rather than launching another reporting campaign. Choose a unit, warehouse, maintenance group, or contractor interface where exposure is real enough to test the system.

For 30 days, compare five sources: official safety reports, supervisor shift notes, medical or first-aid contacts, maintenance defects, and permit or audit corrections. Then interview a small cross-section of workers and supervisors with one practical question: what would make a person hesitate before reporting a weak signal here?

The answer will usually be specific. People may fear blame, paperwork, peer judgment, lost incentives, production delay, contractor penalties, or the belief that nothing changes. Each reason points to a management correction, not a communication slogan.

Connect the findings to executive safety dashboard metrics. Leaders should see not only how many events were reported, but whether the reporting system is credible enough to reveal weak signals before harm occurs.

The most dangerous dashboard is not the one with bad numbers. It is the one that looks clean because workers have learned that the truth is expensive.

Conclusion

Underreporting in safety turns the dashboard into a comfort document. It protects leaders from discomfort while leaving workers exposed to weak controls, unspoken pain, and precursor events that never reach decision-makers.

Safety is about coming home, and that requires metrics honest enough to disturb leadership before harm does. For organizations ready to test whether their clean numbers reflect real control, Andreza Araujo and ACS Global Ventures can support a practical safety culture diagnostic at Andreza Araujo.