6 Decisions from Episode 11 with Dr. Thomas Krause

Episode 11 of Headline Podcast, published on December 3, 2025, brought Andreza Araujo and Dr. Megan Tranter into conversation with Dr. Thomas Krause. The central thesis was that safety performance follows the quality of leadership people experience when the work becomes difficult.

Leadership quality in safety is the practical ability to make risk visible, make safe work possible, and respond well when the first plan does not fit reality. It matters because a declared value has little influence unless people can feel it in decisions, resources, questions, and follow-up.

Krause's argument changes the usual starting point. Instead of asking whether a site has the right policy, leaders should ask whether their decisions make the policy usable at 6 a.m., during a production interruption, or when a worker needs to challenge a confident supervisor.

That distinction fits Andreza Araujo's experience across 25+ years of EHS leadership in multinational operations. In more than 250 cultural transformation projects, the recurring test has not been whether leaders can describe safety well. It has been whether their choices reduce the friction that makes safe work harder than unsafe work.

As OSHA explains in its recommended practices for safety and health programs, management leadership and worker participation have to operate together. ISO 45001 also specifies a management-system approach that connects leadership, planning, support, operation, evaluation, and improvement. The podcast conversation gives those requirements a human test.

1. Treat leadership quality as a control condition

Dr. Thomas Krause said on Headline Podcast, "We were surprised to learn the strongest predictor of success was the quality of leadership given to the initiative." That statement is more demanding than a leadership competency model because it treats leadership behavior as part of the control environment, not as a soft attribute that sits beside operational risk.

The practical question is what the leader does when the plan encounters resistance. A strong leader clarifies the decision, protects the person who raised the concern, and changes the work when the control cannot be executed. A weak leader repeats the value statement and waits for the frontline to absorb the conflict.

Leaders can test this condition in 30 days by sampling 10 recent risk decisions. For each one, review who made the call, what evidence was available, which trade-off was accepted, and whether the decision was revisited after new information appeared.

Andreza Araujo's work links this discipline to measurable outcomes. During a 180-day plan at PepsiCo South America Foods, the accident ratio fell 50% in six months. The lesson is not that every operation can copy the number. The lesson is that leadership becomes credible when it changes the cadence and ownership of decisions.

2. Make safety value felt before asking people to repeat it

Krause's second idea is that value has to be felt, not merely declared. A poster can state that safety comes first, but the workforce reads value through overtime approvals, maintenance priorities, staffing decisions, and what happens after a person says that a task is not ready.

A useful review looks at 5 moments where the organization spends or withholds capacity. Ask whether the site funded the control, allowed the time, assigned a competent owner, accepted a slower sequence, and closed the loop with the people affected.

When those answers are consistently positive, the value becomes observable. When they are inconsistent, the workforce learns that safety is conditional on production pressure. That learning spreads quickly because people compare decisions across shifts, departments, and 19-country networks.

OSHA describes worker participation as a source of information about hazards and solutions. Participation becomes credible only when the information changes something. Otherwise, the organization has collected voice without granting it influence.

3. Walk the floor with questions that expose the gap

Krause's conversation repeatedly returns to the floor because leadership cannot diagnose operational reality from a monthly dashboard alone. A field visit should not be a ceremonial walk. It should be a short investigation into where the written system and the performed job separate.

Use 4 questions. What makes this task harder than the procedure suggests? Which control is most likely to be bypassed today? Who can stop the work without losing credibility? What did the last shift learn that this shift has not yet received?

These questions produce better evidence than asking whether everything is safe, because the answer to a broad question is often shaped by hierarchy. A specific question invites a specific observation, while a respectful follow-up shows whether the leader wants truth or reassurance.

NIOSH explains the hierarchy of controls as an ordered way to reduce exposure. A floor walk should therefore test the control itself, not only whether a person remembers the instruction. If the operator has to improvise around a missing guard, unclear boundary, or unreliable isolation, the leadership question is why the system made improvisation necessary.

4. Investigate why correct work became difficult

Episode 11 challenges a common incident-analysis reflex. Dr. Thomas Krause said, "Incident analysis always looks like it was the employee's fault for not following the procedure, but look deeper and you see following it was made very difficult by system factors set by decisions made a year, or five years, ago." The point is not to remove responsibility. It is to locate the conditions that shaped the choice.

Begin with the task as it existed on the day, then trace the decisions that formed it. Review staffing, equipment selection, maintenance backlog, training assumptions, layout, incentives, previous deviations, and the time pressure that surrounded the work.

A 24-hour evidence freeze can protect the first facts, while a 72-hour review can separate what was observed from what was inferred. The investigation should also identify which decisions were made 1 year earlier and which were inherited from 5 years earlier, because time can hide the origin of a weak control.

Headline has already explored proof and silence in incident investigation. The connection is direct. Leaders lose learning when they collect a neat explanation before they understand the work system that produced the event.

5. Read trust as a distribution, not an average

Krause offered a warning about averages. He said, "You could have 60% of people say they trust their supervisor and still be in the 90th percentile, so the leader pats himself on the back while 40% of his people don't trust their boss." The number is useful because it shows how a favorable score can hide a large group that stays silent.

Trust data should therefore be split by shift, tenure, contract status, location, and relationship to the supervisor. A site with 60% positive responses may have a serious problem if the remaining 40% sit in the night crew or in the group that performs the highest-risk work.

Use 3 follow-up tests. Compare survey scores with near-miss reporting, review whether concerns receive a response within 48 hours, and interview people who chose neutral rather than positive. Neutral answers often carry more operational information than a clean average suggests.

Psychological safety is not a popularity contest. It is the practical capacity to raise a concern before exposure becomes harm, and to know that the response will be serious, specific, and fair.

6. Turn insight into a repeatable leadership cadence

The final decision is to build a cadence that keeps the conversation alive after the podcast ends. Leadership quality cannot depend on an exceptional visit or a single campaign. It needs a rhythm that connects field evidence to decisions within a known time.

A monthly leadership review can use 5 inputs: one field observation, one unresolved concern, one control failure, one incident-learning action, and one decision that changed the work. The point is not to create another meeting. It is to make leadership behavior visible enough to improve.

ISO 45001 frames continual improvement as part of the occupational health and safety system. A cadence gives that phrase operational meaning because every cycle asks what changed, who verified it, and what evidence supports the next decision.

What changes when leadership is felt

When leadership quality improves, the first visible change is not a perfect safety score. It is a shorter distance between a concern and a decision, clearer ownership during uncertainty, and less need for workers to protect themselves through silence or workarounds.

That is why the most useful evidence is operational. Look for a concern that reached the right owner within 48 hours, a control that was funded before work resumed, an incident action that changed the task, and a supervisor who can explain what was learned from the last shift.

These signals are modest, but they are harder to fake than a campaign slogan. They show whether the organization has moved from declaring value to making value usable.

Recommendation

Choose one operational area and run a 30-day leadership-quality review. Sample 10 decisions, split trust data into at least 3 meaningful groups, complete 4 field questions on every visit, and track whether incident actions change task conditions rather than only worker instruction.

Ask the senior leader to sponsor the review, the line manager to own the response, and the EHS professional to test the evidence. That division matters because safety leadership becomes weak when EHS owns every improvement while the operating line owns none of the decision.

The review should end with 3 statements. This is the risk we can now see. This is the decision that will change the work. This is the evidence we will check within 30 days. If the organization cannot complete those sentences, it has an insight problem, not a measurement problem.

The risk is not that a leader misses one perfect question. The risk is that the same unanswered question survives 30 days, 6 months, and another incident review because the organization has mistaken a declared value for a working control.

Headline Podcast exists to make these conversations usable after the microphone is off. Listen to the full conversation with Dr. Thomas Krause, then bring one decision from the episode into your next field review.