Safety Coaching: 5 Decisions That Determine Whether Feedback Changes the Next Task

A supervisor notices the same shortcut during three different shifts. Each conversation ends with a reminder to “be more careful,” yet the task still rewards speed, the equipment still forces an awkward reach, and the production plan still leaves no recovery time. The visible problem looks behavioral. The unresolved problem is managerial.

Safety coaching changes the next task only when five management decisions make the safer action practical, authorized, and observable. Without those decisions, coaching becomes a polished way to transfer design failures to the person closest to the exposure.

Why a coaching conversation is not a control by itself

Coaching can clarify expectations, develop judgment, and give a worker a chance to explain what the task demands. It cannot remove a guarding defect, create enough staffing, correct an impossible sequence, or give a contractor authority that the contract never grants.

James Reason’s work on organizational accidents helps separate the active error from the conditions that make it likely. A person may take the visible shortcut while several latent conditions, including weak design, time pressure, unclear ownership, and unreliable supervision, have already narrowed the available choices.

Andreza Araujo’s Safety Culture: From Theory to Practice gives this distinction a practical leadership test. Culture appears in repeated management decisions, especially when production pressure competes with the control that leaders say matters. A conversation that ignores those decisions may sound supportive while leaving the exposure intact.

The useful question is not whether the supervisor delivered feedback. It is whether the organization changed something that makes the safe behavior easier to perform and harder to bypass.

Decision 1: Is the observed behavior actually the problem to solve?

Observation begins with a visible act, but the visible act is not always the right unit of analysis. A worker who climbs over a barrier may be responding to a blocked route, a missing tool, a poor line of sight, or a work order that makes the approved method impractical.

Before coaching, describe the task as it was performed and compare it with the task as the procedure assumes it will be performed. The comparison should include equipment condition, access, material position, staffing, handover quality, and the time available to complete the work.

If the same behavior appears across people who do not share the same attitude, the organization has stronger evidence of a task or system problem. That evidence does not excuse the action. It tells leaders where a behavior-focused response will be too small.

Ask the worker what the shortcut made possible, what the approved method made difficult, and what would have to change for the safer method to work. Those questions produce better evidence than a warning that simply repeats the rule.

Decision 2: Does the worker have a credible safe option?

Safety coaching assumes that a person can choose the expected behavior at the moment of risk. That assumption fails when the safe option is unavailable, slower without operational allowance, or incompatible with the equipment and work sequence.

A credible option has four properties. The worker can access it, understands how to use it, has enough time to perform it, and knows that the supervisor will protect the decision when output is delayed. If one of these conditions is absent, the conversation should include a management action rather than only a behavioral commitment.

ISO 45001:2018 connects operational control with the conditions in which work is planned and performed. A procedure is not proof of control effectiveness, because the field has to provide the resources, competence, and conditions that the procedure requires.

Leaders should test the option during normal work and during a foreseeable abnormal condition. The second test matters because many controls work only while the task remains predictable.

Decision 3: What authority does the conversation give back to the frontline?

Feedback without authority creates a narrow form of accountability. The worker is told what to do, but the person may still be unable to pause the job, request a replacement, reject a defective tool, or escalate a conflict between the procedure and the schedule.

Every coaching discussion about a safety-critical behavior should identify the decision the worker is allowed to make. That decision may be stopping the task, calling for a supervisor, requesting maintenance, asking for a second person, or refusing to continue until a control is restored.

Authority must be visible in the response that follows. If a worker uses the authority and receives informal punishment, the formal message becomes irrelevant. The next person learns from the consequence, not from the poster or the meeting.

Supervisors can make the authority concrete by documenting the trigger, the first escalation route, the expected response time, and the restart condition. This turns a broad promise to speak up into an operating arrangement that can be checked.

Decision 4: Will the supervisor change the work, or only the worker?

A coaching conversation becomes weak when the supervisor records an individual action and closes the matter without assigning a change to the work. That pattern is common because individual commitments are easy to document, while design changes require coordination, cost, and follow-up.

The supervisor should leave the conversation with two actions. The first belongs to the worker and states the safe behavior that applies now. The second belongs to management and states what will change in the task, equipment, staffing, planning, or supervision.

The management action needs an owner who can make the change, not merely observe the problem. It also needs a verification date and evidence that can be gathered in the field. A completed ticket does not prove that a control became usable.

When the same coaching topic returns, treat recurrence as evidence. Repetition may show a competence gap, but it may also show that the previous management action addressed the record rather than the condition. The review should decide which explanation fits the available evidence.

Decision 5: How will leaders know whether behavior changed under pressure?

A worker can perform the expected action during a planned observation and still revert when the shift is short, the job is late, or the normal tool is unavailable. A single observation therefore provides limited assurance, especially for tasks that change with workload and operating conditions.

Verification should sample the behavior across different shifts, supervisors, task stages, and pressure conditions. The purpose is not to create a larger compliance score. It is to test whether the control survives the conditions that previously made the shortcut attractive.

Choose evidence that reflects the task. It may include the time needed to set up a barrier, the availability of the correct tool, the quality of a handover, the number of repeated exceptions, or the percentage of corrective actions verified in the field. Use a measure only when it helps a leader decide what to change.

Andreza Araujo’s work across more than 250 cultural transformation projects is relevant here only as a leadership principle, not as a substitute for local evidence. Durable change depends on what managers repeatedly protect, verify, and correct after the initial message has faded.

Comparison: behavior coaching versus control improvement

The two approaches can look similar in a meeting record, yet they produce different evidence. Behavior coaching focuses on what the individual should do next. Control improvement also asks what the organization must change so that the expected action remains possible.

What supervisors should change in the next conversation

Start with the task rather than the label. Ask what the worker was trying to accomplish, what the approved method required, and where the two diverged. Then identify the immediate safe action and the management action that will make it sustainable.

Record the authority that applies if the control is missing. Explain how escalation will work, who must respond, and what evidence will confirm that the condition changed. If the worker cannot name a credible route to pause or correct the task, the conversation has not yet restored control.

Review recurrence without humiliation. A repeated shortcut is a signal that deserves analysis, because blaming the person may remove the report while leaving the exposure available to the next shift. Leaders who want honest information must make the response useful and proportionate.

If your organization needs to connect frontline behavior with stronger operational controls, explore Andreza Araujo’s safety leadership resources. For teams building a broader culture program, review the practical ideas behind Safety Culture: From Theory to Practice.

Conclusion

Safety coaching is valuable when it helps a person recognize risk and make a better decision. It becomes a weak substitute for management when the organization leaves the task, authority, resources, and supervision unchanged.

The five decisions provide a practical test. Confirm that the behavior is understood in context, that a safe option exists, that frontline authority is credible, that management will change the work, and that leaders will verify performance under pressure. When those decisions align, coaching can strengthen safe behavior instead of hiding a control failure behind an individual record.