How to Calibrate Safety Coaching Across Supervisors in 14 Days

Two supervisors can watch the same task, see the same deviation, and give opposite advice. One may correct the worker immediately, while the other may ask whether the procedure, equipment, or schedule made the expected behavior possible. Without calibration, safety coaching becomes a matter of personal style, and the next shift receives a different safety standard.

This 14-day process gives a plant manager, operations leader, or EHS coordinator a practical way to align coaching before it becomes another collection of inconsistent observations. The goal is not to make every conversation sound identical. The goal is to make the evidence, decision boundary, and follow-up consistent enough that coaching improves the next task.

What you need before starting

Bring the current safe-work procedures, recent behavioral observations, open corrective actions, task risk assessments, training and competence records, and the names of supervisors who coach the same work group. Select one routine task with enough variation to reveal judgment, such as loading, line breaking, mobile equipment movement, or manual handling.

Use James Reason's systems view as the boundary for the exercise. A visible mistake can be a worker choice, but it can also reveal latent conditions in design, supervision, equipment, workload, or the procedure itself. Andreza Araújo's Guide to Behavioral Observation: ¿VAMOS A HABLAR? adds a practical reminder that observation should open a useful dialogue rather than reduce the person to a score.

Keep the first cycle small. Three to five supervisors, one task family, and one shared evidence sheet are enough to expose the main differences in judgment.

Step 1: Define the coaching decision

Write the decision that supervisors must make after observing the task. It might be whether to correct an immediate exposure, ask a question, stop the job, change a control, or refer a competence gap for formal support. A vague objective such as “improve observations” does not give the group anything to calibrate.

State what good coaching must protect. For a lifting task, the decision may protect the exclusion zone and load path. For manual handling, it may protect the planned method and the worker's ability to request help. The decision should describe the control that matters, not merely the behavior that is easy to notice.

Verify the step by asking each supervisor to describe the same decision in one sentence. If the answers name different outcomes, the calibration target is still too broad. A common error is to begin with a form instead of a decision, which produces more completed records without better judgment.

Step 2: Choose one observable task

Select a task that supervisors can see directly during the 14-day cycle. Do not start with an abstract category such as “safe attitude” or “rule compliance.” Choose a work sequence with a clear start, a meaningful control, and an observable result.

Describe the task in operational terms, including the work area, equipment, main exposure, and control that should be available. The description should be narrow enough that two supervisors can watch the same moment and discuss the same evidence. If the task changes materially between shifts, record the change rather than pretending that the conditions are identical.

Verify that the selected task appears in normal work during the cycle. A common error is to choose a task that only occurs during a special audit, because the team then calibrates for inspection behavior rather than routine work.

Step 3: Separate behavior from conditions

Create three evidence columns titled observed behavior, operating condition, and control consequence. The first records what the person did. The second records what shaped the choice, such as access, equipment condition, time pressure, unclear sequencing, or conflicting instructions. The third records what could happen if the control remains weak.

This separation prevents a coaching conversation from turning every deviation into a character judgment. Andreza Araújo's A Ilusão da Conformidade, or The Illusion of Compliance, is relevant here because visible agreement with a rule does not prove that the work system makes the rule executable.

Verify the evidence by removing adjectives from the first column. “Careless” is an interpretation. “Started the transfer before the receiving operator confirmed the exclusion zone” is observable. A common error is to place the proposed solution in the evidence field, which hides the condition that needs to change.

Step 4: Set the immediate-response boundary

Agree on which conditions require immediate intervention. The boundary should include an uncontrolled exposure, a missing or bypassed critical control, a person entering a hazardous zone, and any change that makes the approved method invalid. Supervisors should know when to pause the task before they begin comparing coaching styles.

Also define what can be discussed after the task. A low-consequence variation may need a question, a demonstration, or a later review rather than a public correction. This distinction protects both control reliability and the dignity of the conversation.

Verify the boundary with short scenarios. Give every supervisor the same four situations and ask what they would do first. A common error is to make “stop work” the only agreed response, because that turns coaching into escalation and leaves no useful route for ordinary skill-building.

Step 5: Run paired observations

Pair supervisors and have them observe the same task without comparing notes during the observation. Each person records the behavior, the condition, the control consequence, and the first coaching response independently. The point is to reveal how judgment differs before the group tries to harmonize it.

Use the behavioral observation conversation method as a companion, but keep this exercise focused on calibration. The pair should not turn the worker into a training subject or create a staged performance. Normal work gives the most useful evidence.

Verify the step with at least three paired observations across different times or crews. A common error is to pair supervisors who already agree on every issue, which creates comfort but reveals little about the decision boundary.

Step 6: Compare evidence before opinions

Bring the pairs together and compare the written evidence before discussing whether the worker was right or wrong. Identify where supervisors recorded different facts, different conditions, or different consequences. Differences in observation often explain differences in coaching better than differences in attitude.

Use a simple rule for the discussion. First agree on what was visible. Then agree on which condition influenced the behavior. Finally decide whether the control consequence requires immediate action, coaching, or system correction. This sequence keeps the group from defending a preferred answer before the evidence is stable.

Andreza Araújo's work on safety culture repeatedly connects maturity with repeated operating choices. In more than 25 years leading EHS in multinational operations, her position has remained practical: a coaching system is credible only when the expected decision can be reproduced under ordinary pressure.

Verify the step by recording two or three examples where the group changed its initial interpretation after examining the conditions. A common error is to treat disagreement as failure. Useful disagreement shows exactly where the calibration rule needs to become clearer.

Step 7: Rehearse the coaching conversation

Ask each supervisor to practice a short conversation using the same evidence. The opening should describe what was seen, invite the worker's explanation, identify the control that matters, and agree on the next safe action. The supervisor should be able to challenge the method without assuming intent.

Keep the conversation proportional to the risk. A critical-control failure needs a firm interruption and a clear restart condition. A technique difference may need a question and a demonstration. A competence gap needs a route to supervised practice or formal training, because a conversation alone cannot create skill.

Verify the rehearsal with a peer checklist. The supervisor should describe evidence rather than labels, ask about conditions, state the control consequence, and close with a specific next action. A common error is to praise the worker for “being safe” without naming the control that should be repeated.

Step 8: Verify the next task

Return to the same task within the final days of the cycle and check whether the agreed coaching changed the next execution. Look for control availability, method clarity, supervisor consistency, and the worker's ability to explain what changed. The purpose is not to catch the worker. It is to test whether coaching reached the work.

Record what improved and what did not. If the same deviation returns because equipment, staffing, layout, or schedule remains unchanged, escalate the condition instead of repeating the conversation. James Reason's model is useful here because recurrence can signal a latent failure that individual reminders cannot remove.

Close the cycle with the five decisions that make coaching change the next task. Andreza Araújo's contribution is not a promise that every conversation will produce immediate compliance. It is the insistence that safer behavior must be made practical through engineering, clear leadership, and care for the person doing the work.

What the 14-day calibration should produce

At the end of the cycle, supervisors should share a short decision guide, a small set of paired examples, and a follow-up rule for recurring conditions. The guide should explain when to intervene, when to coach, when to refer a competence issue, and when to escalate a work-design problem.

• One defined task and one observable control.
• A shared distinction between behavior, condition, and consequence.
• Agreed immediate-response boundaries for uncontrolled exposure.
• Paired observation examples that supervisors can discuss without blame.
• A verification date that tests whether the next task changed.

Calibration does not mean removing supervisor judgment. It makes judgment more dependable by giving it common evidence and a clear decision boundary. When supervisors can explain why they intervened, what condition they found, and what control must hold next, safety coaching becomes part of operational control rather than a separate performance ritual.