How to Run a Weekly Safety Decision Review When the Plan and Field Do Not Match

F2 how-to guide for plant managers, operations leaders, and EHS managers

A production plan can be approved on Monday and become unsafe by Thursday. The shift may inherit a changed task, a missing person, a temporary repair, or a control that exists on paper but is no longer working in the field.

A weekly safety decision review gives leaders a disciplined way to compare the plan with operating evidence. It is not another status meeting. The purpose is to decide what must change, who owns the decision, and when the evidence will be checked again. This method complements the management-system logic of ISO 45001:2018 and the risk-governance principles of ISO 31000.

What you need before starting

Bring the weekly production plan, the current risk register, open corrective actions, recent near-miss and incident reports, control-verification records, staffing changes, maintenance constraints, and any temporary-control register. If the site uses a risk register and decision log, bring both, because the review must connect exposure with an actual management choice.

Invite the operations leader, the EHS lead, the maintenance representative, and the supervisor who can describe the work as it is performed. The group should have authority to pause, redesign, resource, or escalate work. A meeting that can only record concerns will create a longer list without changing the exposure.

Step 1: State the decision the review must make

Open with one decision question, such as whether a high-risk task can proceed under the current controls, whether a temporary arrangement can remain in place, or whether production needs a different sequence. Write the question in plain language before anyone presents a dashboard.

The question matters because a safety review can become a tour of information. When the decision is explicit, participants know which evidence is relevant and which detail can wait. The chair should name the decision owner, the people who provide evidence, and the point at which unresolved disagreement must be escalated.

Verify the step. Ask each participant to complete the sentence, “By the end of this review, we will decide whether…” If the answers differ, rewrite the question.

Common error. Starting with last week's injury rate, because a familiar number can absorb attention without answering the operational question.

Step 2: Compare planned work with actual work

Place the approved plan beside field evidence from the week. Compare work scope, equipment condition, staffing, contractor interfaces, production changes, and the controls that were expected to be present. The point is not to punish variance. It is to identify where a decision was made using conditions that no longer exist.

A supervisor may report that a task followed the procedure while the field record shows a different access route, a changed isolation boundary, or a workaround that became routine. James Reason's work on latent failures is useful here because the visible deviation may be the final expression of an earlier design or management decision.

Verify the step. Select one material difference and ask which control, owner, and decision changed with it.

Common error. Treating the approved plan as the source of truth when the people performing the work describe a different sequence.

Step 3: Sort evidence by decision value

Separate evidence into four groups. The first group shows exposure, such as high-energy work, simultaneous operations, or an unplanned access route. The second shows control condition, such as a failed interlock, overdue inspection, or incomplete isolation. The third shows decision quality, such as a rejected stop-work request, a delayed escalation, or an approval made without the required technical input. The fourth shows follow-through, including actions that remain open or were closed without field verification.

This sorting prevents the meeting from treating every observation as equally important. A low-severity housekeeping observation can remain useful, although it should not displace a control failure that can expose people to serious harm. The comparison between near-miss rate, corrective-action aging, and control verification offers a helpful way to keep evidence tied to the decision rather than to the most visible count.

Verify the step. For each item, record the decision it could change. If no decision changes, move the item to the routine work queue.

Common error. Giving equal meeting time to evidence that describes activity and evidence that tests whether a critical control can be trusted.

Step 4: Test the control, not the document

Choose the control that is supposed to prevent or limit the exposure and ask how the team knows it works today. A procedure, training record, or signed inspection can support the review, yet none of them proves that the barrier is available, capable, and used under the actual operating conditions.

Use a short field check, a direct interview, or a controlled demonstration that matches the hazard and the equipment. If the control depends on a person noticing, deciding, and acting within a narrow window, the review should test that chain rather than accept a completed form. The result should state whether the control is verified, unverified, degraded, or absent.

Verify the step. Ask the control owner to identify the evidence that would disprove the claim that the barrier is working.

Common error. Closing the action because the document is complete even though the physical condition has not been checked.

Step 5: Check whether production pressure changed the decision

Ask what made the safer option difficult during the week. The answer may be a missed delivery, a staffing gap, a maintenance backlog, a late design change, or a performance target that rewarded speed without recognizing the exposure created by the shortcut.

Do not frame this question as a search for bad intent. Andreza Araujo's *Safety Culture: From Theory to Practice* treats the gap between declared expectations and repeated operating decisions as a leadership problem. If the same compromise appears again, the system is teaching people what the organization will tolerate, even when its written policy says otherwise.

Verify the step. Name the operational condition that made the unsafe option easier and assign an owner to change that condition.

Common error. Telling workers to be more careful when the work design, schedule, or resource decision keeps creating the same pressure.

Step 6: Assign one accountable owner and one escalation path

Give each decision one accountable owner who can change the work, not only report the concern. Name the supporting roles separately, because a long list of participants can hide the absence of authority. Set the escalation path when the owner cannot resolve the exposure within the agreed time.

Use a decision log with the exposure, required action, owner, due date, evidence required for closure, and escalation trigger. The risk-governance questions used at board level can be adapted for a plant review when the decision crosses budget, production, engineering, or staffing boundaries.

Verify the step. Ask the named owner what authority, resource, or approval is needed to complete the action.

Common error. Assigning an action to the EHS team when operations, maintenance, or engineering owns the condition that must change.

Step 7: Decide what happens before the next review

Close each material item with a clear interim decision. The options may include proceeding with verified controls, changing the sequence, adding a temporary barrier, stopping the task, or escalating to a leader with authority to accept or reject the remaining exposure. The decision must state what cannot happen while the evidence is incomplete.

Temporary measures deserve a short expiry date and a named review point. If the same temporary measure appears week after week, the review should treat persistence as evidence that the permanent solution is under-resourced or poorly designed. A temporary control that has no exit condition is becoming part of the process.

Verify the step. Read the decision aloud and ask the supervisor who will apply it on the next shift to explain the boundary.

Common error. Writing “monitor” without naming what will be monitored, who will act, and what result will trigger escalation.

Step 8: Verify the change in the field

Schedule the verification before the meeting ends. Match the verifier to the risk, and ask for evidence that shows the changed condition rather than evidence that only confirms a task was assigned. A photograph, observation, test result, interview, or completed work order may be appropriate, depending on the control.

The next review should begin with unresolved decisions and the evidence collected since the prior meeting. That rhythm creates accountability without turning the meeting into a ceremony. It also gives workers a reason to raise concerns, because they can see whether a reported problem changed the work.

Verify the step. Confirm the verification date, location, evidence type, and person who will decide whether the action is effective.

Common error. Measuring completion by the number of closed actions instead of asking whether the exposure and control condition changed.

What should happen when evidence conflicts?

When a procedure says the control is present but field evidence says it is not reliable, pause the decision that depends on the control and resolve the conflict at the right technical level. When the dashboard improves but control verification weakens, treat the divergence as a governance signal rather than as proof of progress.

The chair should record the conflicting evidence, the interim boundary, and the person responsible for resolving the technical question. A disagreement is not a meeting failure when it makes an unsafe assumption visible. It becomes a failure when the group buries the disagreement to protect the schedule.

How should leaders keep the review useful after the first month?

Keep the same decision logic while rotating the evidence that is tested. Review one planned-versus-actual difference, one critical control, one repeated pressure condition, and one action whose effectiveness was previously accepted. This creates a stable cadence without forcing every week into the same script.

Plant managers can also link the review to the site's shift-handover safety review, because unresolved decisions often lose clarity when responsibility moves between teams. The weekly review should not replace frontline conversations. It should make the decisions from those conversations visible and owned.

For a broader leadership reference, Andreza Araujo's *Antifragile Leadership* develops the idea that a stronger organization uses pressure and disruption to improve its decisions rather than hiding the signals that reveal weakness. The practical test is simple. After several weeks, can the team show that difficult evidence changed a resource, a sequence, a control, or an escalation?

Weekly decision-review checklist

• State one decision question and name the accountable decision owner.
• Compare the approved plan with actual work, staffing, equipment, and interfaces.
• Separate exposure evidence, control evidence, decision evidence, and follow-through evidence.
• Test one critical control in the field instead of accepting paperwork as proof.
• Record the pressure condition that could make a shortcut attractive.
• Set the interim boundary, escalation trigger, verification date, and evidence type.
• Start the next review with unresolved decisions and verified changes.

When the plan and the field do not match, leadership has a decision problem before it has a reporting problem. A weekly review earns its place when it converts that mismatch into a named owner, a defined boundary, and field evidence that can confirm or reject the decision.

Explore Andreza Araujo's work on safety leadership and culture for more practical guidance on turning safety expectations into operating decisions.