Executive Safety Metrics: 5 Traps That Reward the Wrong Signal

A clean dashboard can still reward the wrong behavior. When the board sees green, it often assumes the site is safe, even when the number was built to flatter itself.

On a recent Headline Podcast conversation, Dr. Thomas Krause noted that an organization can sit in the 90th percentile while only 60% of workers trust their supervisor. That gap matters because people who do not trust the line of command stop feeding it the bad news it needs.

This article is for directors, plant leaders, and senior EHS teams who want safety metrics that change decisions instead of decorating a slide. The core argument is simple. A safety number is useful only when it helps a leader choose a different action tomorrow.

Key Takeaways

• Safety metrics are useful only when they change a decision, not when they make a slide look tidy.
• Zero targets, lagging counts, percentiles, bonuses, and raw volume can all reward the wrong behavior if leaders do not test the signal behind the number.
• In Muito Alem do Zero, Andreza Araujo argues that lagging indicators only show consequence, not cause.
• On Headline Podcast, Dr. Thomas Krause showed how a site can look strong on paper while trust in the supervisor is still weak.
• A safer board pack combines exposure, leading indicators, metric quality, action closeout, and voice measures.

Why a good number is not enough

In Muito Alem do Zero, Andreza Araujo argues that lagging indicators only look in the rearview mirror. They show consequence, not cause, which means they can report a stable past while the current work is already drifting.

On Headline Podcast, co-host Dr. Megan Tranter and Andreza keep returning to the same point. People do not change behavior for a dashboard alone. They change it when the dashboard is tied to an owner, a deadline, and a visible field check that proves the risk changed.

A board that keeps asking for green counts gets green counts. A board that asks which decision changed gets a more honest conversation. That is the difference between reporting and control.

What happens when zero becomes the target?

Zero sounds noble, but when it becomes the only score, it begins to compete with truth. The system starts rewarding silence, and the people who report hard news quickly learn that the safer move is to stop reporting at all.

In Sorte ou Capacidade, Andreza Araujo argues that an accident free streak can be luck or underreporting rather than proof of capability. At Farmington, the report marked conditions safe enough for miners to enter while reality was the worst ever. That is what happens when the number matters more than the field.

A target that punishes bad news does not remove risk. It removes the evidence that risk is growing, which is exactly when leaders need the evidence most.

Why lagging counts arrive too late

TRIR, LTIFR, and similar counts tell leaders what already happened. They arrive after the exposure and after the decision point, which is why they can never be the only monthly number a leader sees.

Lagging indicators still matter, but only as consequence trackers. In Muito Alem do Zero, the point is not that these numbers are useless. The point is that they cannot tell you which barrier is fading or which supervisor decision changed the risk path.

A leader who waits for the injury rate to move is looking at the rearview mirror while the work is already drifting. By the time the count changes, the system has usually been weak for weeks.

How percentile rank hides internal trust loss

On Headline Podcast, Dr. Thomas Krause said an organization can be in the 90th percentile while only 60% of workers trust their supervisor. That is a strong result on paper and a weak result in the field.

Percentiles compare you with peers, not with your own voice climate. A site can look average or better outside and still have a serious silence problem inside, especially when people have learned that the easiest path is to keep their head down.

When trust drops, bad news slows down. The dashboard stays clean for a while, which is why leaders must read trust and metrics together instead of treating one as a substitute for the other.

Why bonus-linked metrics suppress reporting

Targets tied to bonuses invite people to protect the number they are paid to protect. That is how underreporting begins, because the system quietly tells workers what matters more than the truth.

Andreza Araujo makes this point repeatedly in A Ilusao da Conformidade. Good indicators do not guarantee good practices, because people quickly learn what the system rewards and what it punishes.

If the board wants better learning, it should reward verified closeout, field proof, and control health. It should not reward a polished month end count that hides the hard conversations and the unfinished work.

Why raw volume can still be shallow

More reports can be a sign of health, but only when the reports carry new information. A large volume of near miss reports can still be shallow if every item says the same thing and changes nothing.

On Headline Podcast, Michael Emery said coaching culture has no neat KPI, because the real test is whether workers see the safety lead as a collaborative partner or as a compliance policeman. That distinction matters more than the raw count.

A metric is weak when it measures activity without proving that the activity changed a control, a design, or a supervisor decision. Volume alone can give leaders the illusion of learning while the underlying risk stays untouched.

What should leaders ask instead?

The useful question is not how many reports the site collected. It is which decision the data changed, who owns the next move, and whether the field now looks different because of it.

Across more than 250 cultural transformation projects supported by Andreza Araujo, the pattern is consistent. Leaders get the behavior they ask their metric system to buy. If the pack rewards silence, silence grows. If the pack rewards verification, verification grows.

The board pack should therefore test five things at once. It should show exposure, show a leading indicator, show whether the metric itself is still fresh, show whether the corrective action changed the field, and show whether the workforce still trusts the chain of escalation.

A monthly metric stack that resists distortion

A safer monthly pack usually combines one exposure measure, one leading indicator, one metric quality check, one action closure measure, and one voice measure. That mix keeps the dashboard honest because no single number can pretend to explain the whole picture.

If the exposure is rising, the leader sees it. If the reporting quality is fading, the leader sees that too. If a corrective action looked closed but did not change the field, the dashboard should expose the mismatch before the board starts celebrating.

If you want a companion discussion, compare this article with Control Health Metrics: 5 Blind Spots Boards Miss and then listen to Headline Podcast for the conversation that keeps the topic grounded in real decisions.

FAQ

What makes a safety metric useful to a board?

A board useful metric changes a decision. It shows exposure, trend, ownership, and whether the action changed the field, not just the slide.

Is zero a bad target?

Zero is a moral aim, but when it becomes the only score, people learn to protect the number. The metric then starts competing with the truth.

Why do percentiles mislead leaders?

Percentiles compare you with peers, not with your own trust gap, reporting quality, or field proof. A high percentile can hide internal silence.

What should replace TRIR as the only monthly number?

Nothing should replace it one to one. Leaders need a small stack that combines exposure, leading indicators, action closeout, and voice measures.

Conclusion

Safety metrics are not valuable because they are precise. They are valuable because they move a leader toward a better choice. When the number rewards truth, the site gets learning. When it rewards cosmetics, the site gets silence.

Headline Podcast keeps returning to the same point with Dr. Megan Tranter and Andreza Araujo. The best dashboard is the one that changes what happens in the field on the next shift.