How 250+ Safety Culture Projects Changed What Leaders Count as Improvement

A safety dashboard can look healthier while the work remains just as difficult to control. That contradiction appeared repeatedly across more than 250 cultural transformation projects supported by Andreza Araujo.

Teams reported more training, more observations, and more closed actions, yet field evidence did not always show a stronger barrier or a clearer decision. The lesson was not to stop measuring activity. It was to stop confusing activity with improvement.

Initial scenario: the dashboard was full, but the decision was unclear

The organizations in this portfolio were not short of commitment. Leaders invested in campaigns, supervisor routines, audits, action trackers, and local improvement plans. Their reports contained enough positive movement to create confidence, especially when completion rates rose and overdue actions fell.

The difficulty appeared when a leader asked which task became safer because of the activity. A training percentage could not answer it. Neither could the number of observations or actions closed by the due date. Those measures described effort, but they did not yet describe control.

James Reason's work on latent failures helps explain why this distinction matters. A visible deviation may be the last expression of a weakness that began in design, maintenance, supervision, planning, or communication.

The decision: define improvement through changed conditions

The portfolio shifted when leaders agreed to treat improvement as a change in the operating condition rather than a completed activity. Instead of asking whether the team delivered the plan, leaders asked which exposure was reduced and how the organization knew.

This did not create one universal metric. A guarding change, contractor handover, confined-space rescue arrangement, and supervisor escalation routine need different evidence. The common requirement was a visible chain from problem to decision, from decision to changed work, and from changed work to verification.

Andreza Araujo describes this orientation in Safety Culture: From Theory to Practice, where culture becomes credible through choices people can observe in ordinary work.

Execution: replace activity counts with an evidence chain

The teams kept activity measures, but added four questions to each material improvement. What was the starting condition? What decision changed it? What should be different in the task? What evidence shows that the difference remains real?

A recurring hand injury did not produce a satisfactory improvement claim because a refresher briefing was delivered. The review had to show whether the tool, hand position, material flow, guarding, or supervision changed. If the operator still had to place a hand inside the same line of fire, the activity had not solved the decision problem.

The same logic applied to action closure. A corrective action was not treated as effective because a procedure was revised or a purchase order issued. The accountable leader had to verify the control where the task occurred, including a late shift, contractor handover, maintenance interruption, or production restart.

Measured result: success became harder to fake

The measured result across the portfolio was not one accident-rate number. The change was visible in the quality of management decisions. Leaders began separating completed activity from verified control, which made apparent progress less comfortable but more useful.

A report could show that training completion was high while task verification remained weak. It could show that action closure improved while recurrence evidence was missing. It could show that observations increased because reporting became easier, without claiming that exposure automatically fell.

What the case revealed about leadership behavior

Leaders shape measurement through the questions they reward. When executives ask only for completion, teams become skilled at producing completion. When executives ask what changed in the work and where it was verified, the review follows the real control.

EHS credibility improves when technical challenge is connected to operational ownership. A safety professional can identify a weak barrier, but the line leader usually controls the sequence, resources, staffing, authorization, or design needed to strengthen it.

Uncomfortable evidence is often more valuable than a green dashboard. A finding that says the control is not reliable gives leaders a decision. A high activity count without field context gives reassurance without direction.

What to apply in a monthly executive review

A plant manager or regional EHS director can use this case logic without replacing the existing dashboard. Select three material improvements and require each owner to present the starting condition, the decision, the changed work, and the verification evidence.

Ask whether the control was tested during the task, which operating variation could weaken it, and what would cause the team to reopen the action. These questions turn a review from a performance recital into a decision forum.

The critical-control verification review offers a related way to examine where dashboard confidence separates from field evidence.

Where leaders still misread improvement

A lower number of reported concerns is not automatically proof that conditions improved. The number may fall because reporting became harder, supervisors stopped asking, or workers learned that escalation creates delay.

A new procedure is not automatically a new control. A document can clarify intent while the task still depends on memory, improvisation, or a workaround. The procedure matters when it changes authorization, equipment, sequence, supervision, or the worker's available choice.

Conclusion: improvement is a field claim

Across more than 250 safety culture transformation projects, the durable lesson was demanding. Leaders should count activities because activities show effort, yet they should call something an improvement only after the work, barrier, or decision boundary has changed and the new condition has been verified.

Safety performance becomes more trustworthy when the report explains not only what the organization did, but what became different for the person doing the work.

FAQ

What is the difference between a safety activity and a safety improvement?

A safety activity is something the organization did, such as deliver training, complete an observation, or close an action. A safety improvement is a verified change that reduces exposure, strengthens a control, or makes a safer decision more likely during real work.

How did more than 250 safety culture projects change measurement?

They showed that activity counts become misleading when they are separated from field conditions. A stronger review connects the activity to ownership, changed work, and evidence that the control remained usable when production pressure returned.

What should an executive ask before accepting an improvement claim?

Ask what changed in the work, which exposure or decision boundary changed, who owns the control, and where the new condition was verified. A completed training record or closed action is incomplete evidence.

Can leading indicators prove that safety performance improved?

They can support the conclusion, but they do not prove it alone. The indicator needs a clear decision link and field evidence showing that the underlying barrier, task, or work design changed.

Who should own the review of safety improvement?

The operational leader who controls the work should own the decision, while EHS provides technical interpretation and assurance.