Leading Safety Indicators: 5 Myths That Make Prevention Look Measurable

A monthly safety dashboard can look busy while telling leaders almost nothing about whether serious exposure is becoming less likely. The problem is not a lack of data. It is the belief that any activity close to prevention deserves to be called a leading indicator.

Leading indicators are useful only when they test a decision, a control, or a change in exposure. When they count activity without checking effect, they make prevention look measurable while leaving the real risk untouched. This article examines five myths that safety leaders should challenge before the next performance review.

Why leading indicators are easy to confuse with activity counts

A leading indicator should provide evidence about conditions that precede an unwanted outcome. That does not mean every early event is useful. A completed inspection, a closed action, or a training attendance record can sit upstream from harm while remaining disconnected from the barrier that matters.

Andreza Araujo has spent more than 25 years in global EHS leadership, and her work repeatedly connects measurement to the decisions people can make at the point of work. In more than 250 cultural transformation projects supported by Andreza Araujo, the practical question is not how much activity occurred. It is whether the activity changed the conditions under which people perform the task.

That distinction also appears in Safety Culture: From Theory to Practice, where Andreza Araujo treats culture as something visible in choices, routines, and consequences rather than in declarations. A dashboard becomes credible when its measures help an operations leader decide what must change next.

Myth 1: More leading indicators always create better control

More measures can create the appearance of discipline while weakening attention. When a dashboard contains thirty indicators, managers often protect the reporting cycle instead of examining the few conditions that carry the highest consequence.

The trap is especially strong when each department adds its preferred measure. The result is a catalogue of inspections, observations, meetings, training records, action closures, and audit scores whose combined meaning is unclear. A crowded dashboard can hide the absence of one essential question, which is whether the critical control was available and used when exposure was present.

Choose a small set of measures that answer different management questions. One can test whether high-risk work was authorized correctly. Another can test whether the control survived field conditions. A third can test whether overdue weaknesses were escalated before work continued. The set should be small enough that a plant manager can discuss every exception without delegating the meaning to a spreadsheet.

Myth 2: A completed action is proof that risk was reduced

Action closure is evidence of completion, not evidence of effect. A procedure may be revised, a guard may be installed, or a training session may be delivered while the exposure remains unchanged because the control was poorly designed, unavailable during the shift, or bypassed under production pressure.

Safety leaders should separate four events that are often collapsed into one status field. The action was assigned, the action was completed, the control was present, and the control changed the decision or exposure. Those events may occur weeks apart, and sometimes the fourth event never occurs.

Andreza Araujo's book The Illusion of Compliance is useful here because it names the gap between satisfying a requirement and changing the work. Treat the closed action as a prompt for verification. Ask who checked the control, under what operating conditions, and what evidence shows that the original exposure is different.

Myth 3: Training completion is a leading indicator of competence

Training completion proves that a person was registered, attended, or passed a defined step in the learning process. It does not prove that the person can recognize changing exposure, select the right control, or refuse an authorization that no longer matches the task.

This matters when training is used as the default response to every event. A weak procedure, missing equipment, an impossible schedule, and an unclear decision right cannot be repaired by adding another module. Training may be part of the control, but it is rarely the whole control.

Measure competence closer to work. Observe whether the worker can explain the critical hazard, identify the condition that invalidates the plan, and escalate when the control is absent. The observation should be tied to a real task and should lead to a decision about coaching, redesign, supervision, or authorization. If the measure never changes one of those decisions, it is a learning record rather than a leading indicator.

Myth 4: A high observation count means safer behavior

Observation volume is easy to increase because it rewards activity that can be scheduled. It becomes misleading when observers choose easy tasks, record low-consequence findings, or learn that difficult conversations are unwelcome.

The number of observations says little about the quality of the sample. A site can record hundreds of observations without visiting the night shift, a contractor interface, a non-routine maintenance job, or the point where a critical control is most likely to degrade.

A useful observation measure describes the exposure sampled, the control tested, the conversation held, and the decision that followed. It should also reveal what was not sampled. Andreza Araujo's Guide to Behavioral Observation: ¿VAMOS A HABLAR? supports a dialogue-driven approach in which observation is not a scorecard for worker virtue. It is a way to understand how the task is actually being performed and what the system makes difficult.

Myth 5: A rising reporting rate proves that safety performance is getting worse

A rising reporting rate can indicate deteriorating conditions, stronger trust, easier access to the reporting route, or a more visible response from leaders. The count cannot explain which interpretation is correct by itself.

Leaders create confusion when they celebrate an increase one month and criticize it the next. People then learn that the safest reporting strategy is to wait, edit the concern, or solve the problem privately. A measure that punishes bad news eventually becomes a measure of silence.

Pair reporting volume with response quality. Review how quickly concerns received acknowledgement, whether an accountable owner was named, whether the work changed, and whether the same condition returned. The related guide on safety dashboard blind spots develops this distinction between attractive data and decision evidence.

What to do before the next safety review

Start by listing the serious exposures that the operation cannot afford to miss. For each exposure, name the critical control, the decision owner, the condition that invalidates the plan, and the evidence that would show the control worked. Only then select a measure.

Remove measures that report activity without testing a control or decision. Replace them with questions that a supervisor, operations manager, or EHS leader can answer from evidence. The safety data steward guide offers a practical route for making indicator definitions, ownership, and review cadence more reliable.

Andreza Araujo's experience in multinational operations points to a simple test. If a measure rises or falls and nobody changes a priority, authorization, resource, design, or conversation, the measure is not leading the work. It is describing the reporting system.

For deeper work on measurement, culture, and operational leadership, explore Andreza Araujo's books and learning resources at the Andreza Araujo store. Prevention becomes measurable when the organization can show which decision changed before exposure became harm.