Leading Indicators: 4 Interpretations That Should Change the Safety Decision

A safety dashboard can show improvement while the most important control in the operation is becoming less dependable. The problem is rarely arithmetic. It appears when a leading indicator is treated as proof of prevention before anyone asks what changed in the work.

That distinction matters for EHS managers, plant leaders, and supervisors who have to decide whether to continue a task, fund a correction, escalate an exposure, or accept a temporary condition. A leading indicator should shorten the distance between a weak signal and a better decision. When it only rewards activity, it creates a polished record of uncertainty.

Why the same number can tell four different stories

Numbers do not arrive with their interpretation attached. A monthly count of control verifications may rise because supervisors are checking more critical barriers, because an audit campaign increased activity, because the operation added higher-risk work, or because the system finally began recording work that had always occurred. Those explanations lead to different decisions.

ISO 45001:2018 expects organizations to monitor, measure, analyze, and evaluate occupational health and safety performance. That requirement does not mean that every measure is equally meaningful. It means the organization should know what the measure represents, where its limits are, and how the result informs improvement.

James Reason's work on latent failures also supports a cautious reading. A visible metric can reflect planning assumptions, resource decisions, supervision, equipment condition, and reporting design that were set earlier. The dashboard is therefore an entry point for inquiry, not the final answer.

Before reviewing the four interpretations below, ask one practical question. If this number changes next month, which decision should change with it? If nobody can answer, the indicator is not yet connected to management action.

1. A rising indicator may mean better visibility, not safer work

Increased reporting is often welcomed, and it should be when the organization has been missing weak signals. More near-miss reports, field observations, or verification records can show that workers and supervisors are willing to make exposure visible. That is a positive cultural signal, but it is not evidence that the underlying hazard has been reduced.

The second question is what happened after the signal appeared. Did a control change? Did an owner receive the issue within a defined response time? Did the operation remove a recurring condition, or did it simply close records with reminders and retraining? A rising count with no change in exposure may reveal an organization that has improved its listening without improving its protection.

A plant manager can test the interpretation by sampling a small set of recent records. Compare the original condition, the assigned owner, the action taken, the evidence of completion, and the field condition after closure. This is more informative than celebrating a higher total on the monthly slide.

Andreza Araújo's *Safety Culture: From Theory to Practice* is relevant here because it treats culture as something expressed through repeated decisions, not as a favorable statement of intent. Reporting is valuable when the response teaches the organization that inconvenient information can produce a visible change in work.

2. A rising indicator may mean exposure increased faster than control capacity

Some indicators rise because the operation has become more demanding. Contractor hours may expand, maintenance may move into a shutdown window, production may add simultaneous operations, or a new process may create unfamiliar interfaces. If the number of risk observations rises during that period, the correct interpretation may be that exposure grew faster than the organization's ability to control it.

This is where denominator discipline matters. A raw count can hide a change in work volume, task mix, shift pattern, or contractor population. The comparison should preserve enough context to show what was being done, by whom, under which conditions, and with which critical controls.

Leaders should pair the activity measure with a capacity measure. Examples include the percentage of critical tasks with a competent supervisor present, the proportion of control verifications completed before the task begins, the age of unresolved high-consequence actions, and the number of exceptions that require risk acceptance. The purpose is not to create a larger dashboard. It is to expose when demand is outrunning control capacity.

The existing discussion of thresholds that make critical-risk exposure look acceptable offers a useful companion lens. A threshold should prompt a decision about capacity, not merely change the color of a cell.

3. A stable indicator may conceal a changing control condition

Stable numbers feel reassuring because they reduce disagreement. Yet a stable result can coexist with changed equipment, altered staffing, deferred maintenance, new contractors, or a work method that no longer matches the original risk assessment. The measurement process may be consistent while the exposure has moved.

Critical controls need a different verification question than routine activity. Do not ask only whether the inspection occurred. Ask whether the barrier was available, suitable, understood, used, and able to perform under the conditions that existed that day. A completed form can confirm that a form was completed. It cannot, by itself, confirm that the barrier carried the risk.

For a stable indicator to deserve confidence, the organization should define the conditions that make periods comparable. Those conditions may include task population, control standard, operating mode, workforce competence, maintenance status, and the treatment of exceptions. When one of them changes, the dashboard should show the change rather than preserve a false trend line.

This is also why the six checks for testing a safety indicator matter before a measure reaches the executive dashboard. Measurement quality is part of risk control because leaders make resource and operating decisions from the evidence they receive.

4. A falling indicator may signal silence, access failure, or lower exposure

A falling number can be good news, but the organization should earn that interpretation. Fewer reports may mean fewer hazards, better controls, lower work volume, or a completed project. It may also mean that workers stopped using the channel, supervisors stopped recording exceptions, the system became difficult to access, or people learned that raising a concern creates delay without support.

The first test is participation. Who contributed to the measure before and after the decline? Did the mix of shifts, departments, contractors, or supervisors change? Did the definition change? Did the organization close a reporting route, remove time from the pre-job discussion, or change the review cadence?

The second test is independent evidence. Compare the falling signal with maintenance backlog, control deviations, overtime, schedule changes, quality events, stop-work decisions, and worker conversations. No single comparison proves underreporting, but a mismatch tells leaders that the decline needs explanation before it is called improvement.

Andreza's *Safety Culture Diagnosis: Learn how to do your own* is useful for this question because a perception measure becomes stronger when it is tested against observed conditions and management responses. A quiet dashboard should never be allowed to substitute for a credible listening process.

How to connect an indicator to a decision

A dashboard becomes operational when every important indicator has four fields beside it. The first is the decision owner, meaning the person who can change the condition. The second is the response boundary, which defines when the result requires action or escalation. The third is the evidence source, which explains how the organization will test whether the interpretation is credible. The fourth is the review date, which prevents an unresolved concern from becoming permanent background noise.

For example, a rise in control deviations may require a supervisor to pause the task and a maintenance manager to verify equipment condition. A fall in worker reports may require an EHS leader to test access and participation before declaring success. A stable verification rate during a major change may require the project sponsor to revisit whether the original measure still describes the new work.

These links also reduce the temptation to manage the metric instead of the exposure. If the response is defined in advance, leaders can ask whether the action occurred and whether it changed the condition. If there is no response, the organization has a reporting ritual rather than a leading indicator.

What leaders should remove from the monthly review

Remove indicators that reward volume without preserving meaning. A high number of observations is weak when duplicate records, low-quality descriptions, and automatic closures make the count easy to increase. A high completion rate is weak when overdue actions are reclassified, scope is narrowed, or the evidence of effectiveness is never reviewed.

Remove comparisons that hide population changes. A rate may appear favorable after high-risk work is transferred to contractors, after a shutdown ends, or after the operation reduces the number of people exposed without changing the underlying hazard. The metric may be accurate within its definition while still being unsuitable for the decision being made.

Remove the assumption that one indicator can represent culture, control health, and outcome performance at the same time. A dashboard should show how these views relate, but it should not collapse them into one score that cannot explain what leaders need to do next.

How Andreza Araújo's approach keeps metrics close to prevention

Across more than 25 years in multinational EHS leadership, Andreza Araújo has positioned safety metrics as part of a broader operating system that connects culture, leadership, risk perception, and control verification. That position is consistent with her documented PepsiCo South America result, where the accident ratio fell by 50% over six months under a 180-day plan. The lesson is not that one number caused the result. The lesson is that measurement matters when it is attached to a defined change program and visible decisions.

Her book *Make The Difference: Be a Leader in Health & Safety* adds the leadership question that dashboards often omit. Who will make the decision when the metric is inconvenient? A board may receive a green status, while a supervisor is still choosing between a production target and a degraded barrier. The quality of the metric is tested at that point of tension.

For leaders building or revising a monthly review, the practical sequence is clear. Define the exposure, identify the control, select the signal, state the limitation, assign the decision owner, and verify the condition where the work occurs. This sequence creates better evidence than adding another column to an already crowded dashboard.

FAQ: Leading indicators and safety decisions

What makes a leading indicator useful in safety management?

A leading indicator is useful when it provides timely evidence about a condition that leaders can influence and when a defined response changes the work. A count without an owner, threshold, or verification step is activity data rather than dependable prevention evidence.

Does a higher number of safety observations mean performance improved?

Not by itself. A higher count may reflect better participation, more exposure, easier reporting, a campaign, or a change in how observations are classified. Leaders should examine the quality of the observations and whether the related controls changed.

Why can a stable safety dashboard still hide serious risk?

A dashboard can remain stable when the measurement process is stable while the work, equipment, staffing, or control condition changes. Stability becomes meaningful only when the organization verifies that the measured population and the critical controls are still comparable.

How should leaders respond when a leading indicator falls?

First determine whether exposure fell or whether reporting, participation, access, or verification weakened. If the cause is unclear, treat the change as an information gap and assign a short review that tests the signal against field evidence.

Should leading indicators replace injury rates?

No. Injury rates describe outcomes and can support trend review, while leading indicators provide information about conditions and controls before an event. The two serve different purposes, and neither should be treated as a complete picture of risk.

Make the next dashboard review more useful. Explore Andreza Araújo's safety leadership resources and use the book and guide collection to connect metrics with culture, risk perception, and control decisions.

A leading indicator earns its place when it changes what a responsible leader does next. If the number rises, falls, or stays still, the organization should know which interpretation is credible, which evidence will test it, and which decision will follow. That is how a dashboard moves from reporting activity to protecting people.