KPI Theater: 4 Failures That Make Safety Performance Look Better Than It Is

KPI theater starts when a safety dashboard looks disciplined enough to satisfy leadership, but weak enough to miss the risk that is growing in the field. The numbers arrive on time, the colors look stable, the monthly deck has the expected format, and nobody in the meeting asks whether the indicator still represents the work that can injure someone tomorrow.

The problem is not measurement itself. EHS leaders need indicators, trend lines, denominators, and review routines. The problem appears when the organization treats display as control. A clean metric can become a substitute for supervision, field verification, worker voice, and executive decision-making.

Across 25+ years in executive EHS roles, Andreza Araujo has seen the same pattern in different sectors and countries: companies rarely suffer from a shortage of safety numbers. They suffer from indicators whose meaning, owner, threshold, and field connection have decayed. This article diagnoses four failures that make safety performance look better than it is.

Why KPI theater is more dangerous than a bad month

A bad month can be painful, but at least it creates discomfort. KPI theater does the opposite. It gives leaders the emotional reward of control while the system loses sensitivity to weak signals. Because the dashboard is green, the team stops asking why near misses fell, why corrective actions closed faster than evidence improved, or why contractors disappeared from the denominator.

In her Portuguese title *Muito Alem do Zero* ("Far Beyond Zero"), Andreza argues that a rigid zero can protect the number rather than the life. The same logic applies to KPI theater. When the target becomes the object to defend, people learn how to keep the visible indicator clean, even when field conditions deserve a harder conversation.

James Reason's work on latent failures helps explain the danger without blaming the operator. A weak indicator does not usually fail because one person entered a bad number. It fails because incentives, definitions, reporting channels, leadership reactions, and review habits allow distortion to become normal. The number is only the final expression of those conditions.

Failure 1: The KPI has no decision attached to it

The first failure is a KPI that appears every month without naming the decision it should support. The dashboard says training completion is at 98 percent, audit closure is on track, observations increased, or recordable rate decreased. The meeting accepts the number, but nobody can say which work will stop, which exposure will receive resources, which control will be verified, or which leader must act differently.

This is how a metric becomes decorative. It still has a formula, a chart, and a place in the leadership deck, but it no longer changes management behavior. A safety KPI should inform a decision such as escalating a deteriorating trend, checking a critical control, changing a contractor condition, reopening a weak corrective action, or challenging a reporting pattern that looks too clean for the work being performed.

The fix begins with a direct question: what decision does this KPI make better? If the answer is only "monitoring performance," the indicator is too vague for an executive safety meeting. Monitoring can belong in a reference report. Decision indicators deserve the scarce attention of leaders because they name the next action when the signal changes.

The existing safety KPI owner review gives EHS managers a practical way to assign this accountability. It separates the person who extracts the data from the person who owns interpretation and escalation. That distinction matters because a beautiful dashboard without a decision owner still leaves the organization exposed.

Failure 2: The denominator changed while the story stayed the same

The second failure is denominator drift. A rate looks better, but the exposure base changed quietly. Work hours moved to a contractor that is not fully captured. High-risk tasks became seasonal and no longer align with the monthly comparison. Maintenance backlog shifted the risk from routine operations to shutdown work. The numerator may be honest, yet the rate still tells the wrong story.

This is especially dangerous when leaders compare sites. One plant may look weaker because it captures short-duration tasks with discipline, while another looks stronger because it normalizes only by broad hours. A single denominator can flatten very different risk profiles. Total hours may be useful for broad injury rates, but it can be a poor proxy for energy isolation, confined space entry, lifting, hot work, or line-breaking exposure.

Andreza Araujo's work in more than 250 cultural transformation projects reinforces a practical lesson: the field decides whether a metric is credible. If supervisors and workers recognize that the denominator ignores the work that actually worries them, they will treat the dashboard as corporate paperwork rather than risk intelligence.

EHS managers should compare exposure hours, task volume, and control verification before accepting a rate. The article on exposure denominators in safety metrics explains why different decisions require different bases. KPI theater thrives when one denominator is treated as universally objective.

Failure 3: Reporting volume is rewarded without testing signal quality

The third failure is rewarding reporting volume without checking signal quality. A plant can celebrate more observations, more near misses, more inspections, and more corrective actions while the underlying signal becomes weaker. Quantity matters only when the organization also checks whether reports describe real precursors, name usable controls, and lead to decisions that reduce exposure.

Observation counts are a common example. A supervisor can submit many observations that repeat safe conditions, generic behaviors, or minor housekeeping issues. The score improves, but serious exposure remains untouched. Near-miss volume can fail in the opposite direction. A sudden drop may look positive, although it actually signals fear, fatigue, retaliation risk, or a belief that nothing happens after reporting.

As Andreza writes in *A Ilusao da Conformidade* ("The Illusion of Compliance"), good indicators do not guarantee good practices. That point is central to KPI theater. A metric can prove that the form was filled, the meeting occurred, or the action was closed, while saying little about whether the control was effective in the workplace.

The practical test is to audit a sample of signals, not only the count. Pick recent observations, near misses, audit findings, and corrective actions. Ask whether each record names the exposure, the barrier condition, the owner, the deadline, and the verification evidence. The leading indicator quality audit can serve as the method for this review.

Failure 4: Green status is accepted without field contradiction

The fourth failure is accepting green status without field contradiction. A dashboard says the operation is controlled, but field walks show improvised tools, rushed permits, open guards, blocked access, untested emergency response, or procedures that no one can use under pressure. When leadership accepts the dashboard over field evidence, the metric becomes a shield against reality.

Green should never mean "no questions." It should mean the indicator is inside expected variation, the denominator is still valid, reporting channels are open, and field verification did not contradict the score. Without those conditions, green is only a color. It may represent silence, weak sampling, poor classification, or a review culture that punishes bad news.

This is where underreporting becomes a management issue, not an honesty issue. Workers report less when they expect blame, delay, indifference, or paperwork without action. Leaders then see fewer events and treat the absence of reports as proof of control. The existing analysis of underreporting distortions shows how clean safety metrics can hide social pressure.

An EHS manager should require contradiction checks before accepting green in high-risk areas. That means supervisor interviews, task observation, control verification, contractor sampling, and review of weak signals that did not enter the formal dashboard. The article on dashboard blind spots hiding fatal risk covers this problem from the executive review angle.

How to tell performance discipline from KPI theater

The table below separates useful performance discipline from KPI theater. The difference is not whether the company measures. The difference is whether the measurement still creates honest decisions.

What the EHS manager should change this month

The fastest way to break KPI theater is to review the next leadership dashboard before it is presented. Choose the five indicators that receive the most executive attention. For each one, ask four questions: what decision does it support, who owns interpretation, what denominator could distort it, and what field evidence could contradict it?

This review should be uncomfortable in the right places. If a KPI has no decision, move it out of the main meeting. If the owner is a department, name a person. If the denominator does not match the exposure, add a second view. If green cannot survive field verification, change the status or add a note that forces leadership to see the uncertainty.

The goal is not to make the dashboard pessimistic. It is to make the dashboard honest enough to protect people. Safety is about coming home, and a metric that hides risk does not serve that mission, even when it makes the monthly report easier to defend.

For teams that want to connect metrics with culture and leadership routines, Andreza Araujo's Safety School and books give a practical base for moving from reporting to management behavior. The starting point is simple: make every safety KPI answer the decision it exists to improve.

Conclusion: a safety KPI is only useful when it changes a decision

KPI theater is attractive because it looks mature from a distance. It has colors, trends, targets, owners on paper, and enough activity to reassure a busy leadership team. Up close, it often lacks the one thing that matters most: a trustworthy connection between the number and the risk.

A stronger safety dashboard does not need more decoration. It needs fewer passive indicators, better denominators, named decision owners, sampled signal quality, and field contradiction checks. If an indicator cannot influence a real decision, it should not dominate the leadership conversation.

Andreza Araujo's resources at andrezaaraujo.com help leaders connect safety culture, indicators, and field behavior with the seriousness that occupational safety requires. Start by challenging one green KPI this month, because the first honest question often reveals where the dashboard stopped protecting the work.