Safety Coaching: 5 Traps That Keep Supervisors Correcting the Wrong Behavior

A supervisor sees a worker step inside a marked exclusion zone while a load is being positioned. The supervisor points at the line, says “You know better,” and records the observation. The rule has been repeated, yet the next crew may make the same decision before the week ends.

The problem is not that the supervisor corrected the wrong person. The problem is that the conversation stopped before anyone tested the work. Was the load path clear? Could the operator see the signaler? Did the task plan match the layout on the ground? Was the exclusion zone protected when the job changed? A coaching conversation that skips those questions may produce compliance theater while leaving the exposure intact.

Safety coaching is valuable when it turns an observed decision into evidence about the conditions that shaped it. That requires discipline from the supervisor, because the fastest conversation is rarely the most useful one. The five traps below show where well-intended coaching becomes a ritual that corrects people without improving work.

1. Starting with blame instead of the decision

The first trap appears in the opening sentence. “Why did you do that?” often sounds like a question, but its tone carries a conclusion. The worker hears that the supervisor has already decided the behavior was careless, so the safest response becomes a defense, an apology, or silence.

That response gives the supervisor very little evidence. A person may say the procedure was clear, the job was rushed, or the shortcut was a one-time event because each answer seems less risky than describing what actually happened. The conversation then confirms the supervisor’s original belief without testing the decision.

A stronger opening asks the worker to reconstruct the task. “Walk me through what you saw when the load moved.” This wording does not excuse the decision, and it does not remove personal responsibility. It creates enough space to learn whether the person noticed a change in the plan, lost sight of the signal, or believed the exclusion zone had been released.

James Reason’s work on organizational accidents is useful here because it separates the visible action from the conditions that make failure more likely. Andreza Araujo’s *Guide to Behavioral Observation: ¿VAMOS A HABLAR?* reaches the same practical point through dialogue. An observation is not complete when a deviation is named. It becomes useful when the conversation reveals what the team needs to change.

2. Treating the procedure as proof that the task was controlled

Many coaching conversations move directly from an observed action to a procedure. The supervisor quotes the required step, asks for agreement, and closes the record. This approach feels objective because the written instruction is easy to reference, although it assumes that the document describes the work as it was actually performed.

Procedures can be technically correct and operationally weak at the same time. A lifting procedure may require a protected exclusion zone without explaining how the zone will be maintained when vehicles, pedestrians, and temporary storage occupy the same area. A line-breaking procedure may require verification without giving the crew a reliable way to confirm isolation after the plant configuration changes.

The supervisor should therefore compare three things during coaching. The first is the intended control. The second is the condition that existed. The third is the decision that connected them. If those elements do not align, repeating the rule is not a control improvement. It is a reminder that the system already failed to guide the task.

This distinction matters for competence as well. Training completion does not prove that a person can recognize a changing exposure, select the right control, and pause the job when the plan no longer fits. Coaching must test those abilities in the setting where the decision was made, because competence that exists only in a classroom does not protect the next critical task.

3. Assuming repetition means the person is the problem

A repeated deviation is frustrating, especially when the same rule has appeared in briefings, training records, and audit findings. The third trap is concluding that the worker lacks care or discipline before checking why the same behavior remains the easiest option.

Repeated behavior can signal a weak design. The required tool may be stored too far from the task. The permit may be written for the planned sequence even though the crew routinely encounters a different sequence. The supervisor may praise speed while discussing safety as a separate value. The control may also depend on a visual cue that disappears in poor light, noise, dust, or a crowded work area.

When the same deviation appears across people, shifts, or contractors, the evidence points away from an individual explanation. When it appears only with one person, the supervisor still needs to test task conditions before deciding that coaching or retraining is the answer. A person who has not been given a workable method cannot demonstrate reliable compliance by effort alone.

Andreza Araujo’s *Safety Culture: From Theory to Practice* describes culture through the decisions leaders make under operating pressure. That principle changes the supervisor’s question from “How do I make this worker follow the rule?” to “What is the operation rewarding, tolerating, or making difficult in this task?” The second question is harder, but it is more likely to prevent recurrence.

4. Confusing a quiet conversation with psychological safety

A calm tone helps, but a calm tone is not enough. Some supervisors avoid direct questions because they want coaching to feel supportive, so the discussion stays polite while the critical issue remains unnamed. The worker leaves without knowing which decision must change, and the supervisor leaves without knowing whether the person understood the risk.

Psychological safety does not mean removing challenge from the conversation. Amy Edmondson’s research makes the distinction clear. People need enough interpersonal safety to speak honestly, while the team still needs high standards and clear accountability. A supervisor who asks what made the decision difficult, then explains the consequence of leaving the condition unchanged, is doing both.

The conversation should become specific at the point where risk is clarified. The supervisor can ask what signal would trigger a pause, who has authority to stop the task, and what evidence will show that the control is working. Those questions respect the worker as a source of information while refusing to treat explanation as the end of the process.

Coaching becomes performative when the supervisor values a friendly interaction more than a changed decision. Respect is not measured by how comfortable the conversation feels. It is measured by whether people can tell the truth, understand the standard, and leave with a control that can be used under real conditions.

5. Closing the conversation without an owner or verification

The final trap is ending with “Let’s be more careful.” That sentence assigns no action, no owner, and no time for checking whether the task changed. It also places the entire burden on the worker, even when the conversation identified a planning, equipment, supervision, or layout problem.

A credible close states what will change and who will make it happen. The supervisor may need to revise the task sequence, move the storage point, restore a barrier, brief the next shift, or involve engineering. The worker may need to demonstrate the revised method. Both roles can exist in the same plan, because personal attention and system correction are not competing explanations.

Verification should match the risk. A low-consequence housekeeping issue may be checked during the next routine walk. A control protecting people from a suspended load, hazardous energy, or a chemical release deserves an earlier field check whose evidence is visible to the crew. The point is not to create more paperwork. It is to confirm that the agreed change exists where the work happens.

The supervisor should also record what was learned, not only what was observed. A note such as “worker reminded to stay outside the zone” preserves the blame frame. A note such as “zone became unclear after vehicle repositioning; supervisor to mark a second boundary before the next lift” captures a decision that another shift can act on.

What a useful safety-coaching conversation tests

A practical coaching conversation can remain brief without becoming shallow. It tests the relationship between the task, the control, and the decision. The following sequence gives supervisors a usable structure.

• Ask the worker to describe what changed or what they expected to happen.
• Identify the intended control and compare it with the control that was actually available.
• Clarify the decision that was made, including what information was visible at the time.
• Agree on one change for the worker and one change for the work system when both are needed.
• Name the owner, the verification point, and the escalation route if the control cannot be restored.

This structure avoids two weak extremes. It does not reduce the conversation to “follow the rule,” and it does not turn the conversation into an open-ended discussion with no decision. The supervisor keeps the standard visible while investigating whether the operation made that standard practical.

How to tell whether coaching is changing behavior

Coaching quality is visible in the work, not in the number of conversations recorded. A team may complete hundreds of observations while the same exposure remains normal because the observations are counted as activity rather than used as evidence.

Look for changes that another supervisor can see. The task plan may include a condition that was previously ignored. A critical control may have a clear owner. Workers may pause when a boundary becomes unclear because the escalation route is known. The next shift may receive the decision instead of receiving only a reminder to work safely.

Patterns matter more than isolated success. If the same deviation returns after three conversations, the supervisor should reopen the diagnosis rather than repeat the script. The problem may be a missing resource, a production conflict, an unclear standard, or a leadership signal that rewards the shortcut. Repetition is information, especially when the organization has already tried to correct it.

Across 25+ years of multinational EHS leadership, Andreza Araujo has treated safe behavior as a product of engineering, creativity, and care. In more than 250 cultural transformation projects, that view keeps the conversation practical. Leaders must make the safer decision possible, understandable, and expected, then verify that the work supports it.

Conclusion: coach the decision, then change the condition

Safety coaching fails when it treats the observed worker as the whole explanation. Blame closes the evidence. Procedure recitation hides the gap between written intent and field reality. Repetition is misread as personal failure. Politeness replaces clarity. A final reminder is recorded without an owner or a check.

The better approach is disciplined and human. Ask what the worker saw, test which control was available, understand why the decision made sense in that moment, and assign a change that can be verified. If the work system keeps inviting the same shortcut, the supervisor must coach the system as well as the person.

Andreza Araujo’s *Make The Difference: Be a Leader in Health & Safety* offers a useful standard for supervisors who want conversations to matter. Leadership becomes visible when a field observation changes a decision, a resource, or a routine instead of becoming another record in a database.

For more practical conversations about safety leadership, behavior, and risk, explore Headline Podcast and Andreza Araujo’s work at andrezaaraujo.com.