How to Run a Behavioral Observation Conversation in 20 Minutes

F2 how-to guide for supervisors, team leaders, and EHS professionals

A behavioral observation only improves safety when the conversation reveals how work is actually being performed and ends with a decision that makes the next attempt safer. If the observer records a behavior, labels it compliant or non-compliant, and leaves without checking the condition behind it, the form becomes a record rather than a control.

This twenty-minute method helps a supervisor examine one task without turning the exchange into an inspection or a blame exercise. The deliverable is one verified work condition, one practical adjustment, and one follow-up owner.

What you need before starting

Choose a task that is happening today and bring the relevant procedure, the current risk assessment, and the name of the person who can change the work condition. You do not need a long checklist. You need enough context to understand the task, the credible exposure, and the control that should be visible at the point of work.

Set aside twenty minutes and agree that the first part is observation rather than correction. The observer should know the purpose of the visit, while the worker should know that the goal is to improve the work system rather than score personal behavior. That distinction reflects the dialogue method developed in Andreza Araujo's Vamos Falar? guide, where questions are used to expand risk perception instead of closing the conversation too early.

Step 1: Select one task with a meaningful exposure

Start with one task whose conditions can change the level of risk during the shift. Examples include replacing a guarded component, loading a vehicle, handling a chemical transfer, or performing a manual lift with changing access conditions.

Write the task in observable language before arriving. “Safe work” is too broad to guide attention. “Move the solvent container from the storage cage to the mixing point” gives the conversation a defined beginning, sequence, and end.

Verify that the task is active or about to start, because a retrospective discussion often produces a polished description rather than evidence from the work itself. The common error is choosing a convenient route instead of a consequential activity, which turns behavioral observation into a visibility exercise.

Step 2: Define the control you expect to see

State the control in a form that can be seen, demonstrated, or tested. “The worker understands the hazard” is not observable. “The transfer line is isolated, depressurized, and checked before the coupling is opened” can be examined at the point of work.

Use the hierarchy of controls to frame the question. Ask whether the task has been designed to remove the exposure, whether an engineering safeguard is available, and whether an administrative instruction is carrying more weight than it should. PPE can be part of the control, but it should not become the entire definition of safe behavior.

James Reason's work on latent failures supports this discipline because an action at the workface can reflect planning, equipment, supervision, and maintenance decisions made earlier. Verify that the control you named is connected to the exposure, rather than merely being the easiest item to observe.

Step 3: Observe the sequence before asking questions

Watch enough of the task to understand its sequence before beginning the interview. Notice where the worker pauses, reaches, changes position, searches for equipment, asks for help, or creates a workaround when the written method does not match the physical setting.

Do not treat every difference from the procedure as misconduct. A difference can show that the procedure is unclear, the tool is unavailable, the layout creates unnecessary movement, or production timing has changed the task. The point is to identify what condition shaped the action.

Verify your interpretation by recording only concrete observations, such as “the operator moved outside the marked zone to see the gauge,” rather than conclusions such as “the operator ignored the rule.” The common error is judging intent before the sequence is understood.

Step 4: Open with a question about the work

Begin with a question that invites the worker to describe the task in their own order. Ask, “What are you watching most closely right now?” or “Where can this job become more difficult than the written method suggests?”

Listen without interrupting during the first answer. A supervisor who starts by announcing the correct behavior usually receives agreement, whereas a supervisor who asks about the work can learn which condition is creating the gap.

Verify that the conversation stays specific by asking for a physical example. If the worker says the task is difficult, ask which movement, tool, handoff, or timing creates the difficulty. The common error is accepting a general answer that sounds cooperative but cannot guide a change.

Step 5: Separate the action from the condition

Describe the observed action neutrally and then ask what made it reasonable at that moment. This separates the visible behavior from the condition that influenced it without removing personal responsibility for the decision.

For example, “I saw you step around the barrier to reach the valve” is an observation. “What made that route necessary?” is an investigation of the condition. The answer may involve line of sight, equipment placement, congestion, a missing tool, or a work sequence that was never tested with the people who use it.

Verify that the worker can identify a safer alternative that is available now. If the only alternative requires a redesign or a maintenance intervention, record that escalation instead of asking the worker to compensate indefinitely. The common error is praising a workaround because it kept production moving.

Step 6: Test the risk perception behind the choice

Ask the worker what could happen if the condition remained unchanged and which consequence would be most serious. This question tests risk perception without turning the conversation into a quiz about policy language.

Andreza Araujo's 80 Ways to Expand Risk Perception emphasizes that risk becomes actionable when people can connect a hazard with a credible consequence and a decision they control. Use that principle to ask what signal would tell the worker to stop, ask for help, or change the sequence.

Verify that the answer includes a trigger for action, not only a hazard name. “Chemical exposure” is incomplete unless the worker can explain which condition requires the transfer to stop and who can authorize the next step. The common error is mistaking hazard vocabulary for practical awareness.

Step 7: Agree one change that can hold

Choose one change that addresses the condition and can be implemented by a named person. The change might be moving a tool, repairing a guard, revising a task sequence, adding a point-of-use check, or stopping the task until an isolation is verified.

Keep the commitment narrow enough to be completed and important enough to matter. A list of six actions can look thorough while leaving the original exposure unresolved. Andreza Araujo's Safety Culture: From Theory to Practice treats credibility as a product of what leaders make visible and what they actually close, which is why one completed control is more useful than several vague promises.

Verify the owner, due time, and acceptance condition before ending the conversation. “Maintenance will look at it” is not ownership. “The maintenance supervisor will install the fixed barrier before the next batch, and the operator will demonstrate the access route” is testable. The common error is assigning the action to the observer because the observer happened to find it.

Step 8: Close with a follow-up that the worker can see

Repeat the agreed change in plain language and explain when the team will check it. The worker should know what will happen next, while the owner should know what evidence will show that the control is working.

Return to the task at the promised time, even when the change is small. A supervisor who closes the loop teaches that speaking up leads to a visible response. A supervisor who records the observation and never returns teaches the opposite lesson.

Verify the result in the field, not only in the action tracker. Ask the worker whether the change removed the difficulty and whether any new problem appeared. The common error is closing the action because a message was sent rather than because the work condition improved.

Use this final checklist before you leave

A focused behavioral observation is complete when the conversation produces evidence and a decision, not when the form is full. Confirm the following points before leaving the work area.

• Record one task and one observable exposure.
• Describe the action without assigning intent.
• Ask which condition shaped the action.
• Identify the consequence and the stop or escalation trigger.
• Name one owner, one due time, and one acceptance condition.
• Schedule a field follow-up that the worker can see.

Use the difference between habit and behavior when the same action appears repeatedly, and compare the conversation with the stop-work rehearsal method when the exposure requires an explicit intervention. For the broader leadership response, the four response layers of psychological safety help show whether people can raise concerns without being dismissed.

What this method changes

A behavioral observation conversation works when it connects what a person did with the condition that made the choice likely, then converts that understanding into a control that can be checked. The method is not a softer inspection. It is a disciplined way to improve the work without hiding behind labels.

Supervisors who want to strengthen safe behavior can use Andreza Araujo's books and Safety School resources to build a repeatable dialogue practice. Start with one task, one question, and one control that the team can verify before the next shift.