When Safety Culture Moves First: A PepsiCo Case

A 50% reduction in the accident ratio over 6 months is visible in a dashboard, but the dashboard does not explain why the change held. During Andreza Araujo's PepsiCo South America tenure, the important shift happened earlier, when leaders changed the decisions they made under production pressure.

This case study examines that sequence without turning one result into a universal promise. The practical lesson is that safety culture becomes measurable after leaders make the desired behavior easier, more visible, and harder to reverse.

Why the metric moved after the culture changed

Safety culture moved before the accident ratio because leadership behavior changed before the lagging result could register. The 6-month PepsiCo result is best understood as the outcome of repeated decisions, not as the direct product of a campaign or a single training event.

Leaders often ask whether culture can be measured, then reach for a survey score or a total recordable incident rate. Those measures can be useful, although neither one shows what a supervisor does when a schedule slips by 2 hours and a control is inconvenient.

As Andreza Araujo argues in Safety Culture: From Theory to Practice, safety is a value rather than a priority because a priority can be displaced when pressure rises. That distinction gives the case its central thesis. Culture is visible in the trade made at the exact moment when the operation wants speed.

ISO 45001 specifies that organizations should establish processes for eliminating hazards and reducing OH&S risks, and ISO publishes the standard through its official standards system. The case adds an operational question that the standard alone cannot answer: who changes the decision when the control threatens the plan?

1. The starting condition was a leadership problem

The starting condition was not simply a high incident number. It was a gap between the safety message leaders approved and the behavior that the work system rewarded during ordinary production pressure.

That gap matters because employees read consequences faster than slogans. If a supervisor is praised for recovering a delayed order but questioned for stopping an unsafe task, the real value hierarchy is already clear, even when every poster says safety comes first.

Across 25+ years leading EHS in multinational environments, Andreza Araujo identifies this as a coherence test. A culture diagnosis must compare stated values with shift-level decisions, especially the small decisions that never reach the executive dashboard.

For an EHS manager, the first action is to collect 3 evidence layers: what leaders say, what supervisors do, and what the worksite makes possible. The related guide on four evidence layers in safety culture expands that diagnostic sequence.

2. The decision was to change the operating system

The decisive move was to treat safety culture as an operating condition rather than as a communications project. Leaders had to connect expectations with meetings, field presence, escalation, and the allocation of time.

This distinction prevents a common failure. A company can add 4 hours of training, issue a new statement, and still leave the supervisor alone with the same conflicting targets at 5:30 p.m. The message changes, but the decision environment does not.

In more than 250 cultural transformation projects, Andreza Araujo observes that the frontline leader carries the culture into the work. That role cannot be delegated to the EHS department because EHS can advise, challenge, and measure, while line leadership decides what is accepted.

The practical test is simple. When an exception appears, ask which role owns the decision, what evidence is required, and how quickly the issue reaches someone with authority. If those 3 answers are unclear, the organization has a statement rather than an operating system.

3. The first 30 days made leadership behavior visible

The first 30 days should make leadership behavior observable before the organization expects a statistical improvement. A new cadence gives employees evidence that the stated value has entered the work rather than remained in a presentation.

That cadence can include a shift-start risk conversation, a field verification with a named control owner, and a short review of open decisions. The point is not to create 3 more rituals. The point is to expose whether leaders ask about production first or about the conditions that make production safe.

HSE reports that effective health and safety management depends on leadership, worker involvement, and arrangements that are actually implemented. The PepsiCo case illustrates the implementation problem. A procedure can exist for years, while the leader's response to a late job determines whether the procedure has force.

Managers should record 5 observations each week for 4 weeks, focusing on decisions rather than speeches. Review whether controls were protected, whether concerns received an owner, and whether the response reached the person who raised the issue.

4. The 90-day shift connected culture to decisions

By 90 days, a safety culture intervention should be visible in decision records, escalation paths, and supervisor conversations, even if lagging metrics remain noisy. Decision evidence shows whether the new expectation survives contact with work.

A useful record contains the exposure, the decision maker, the control that was protected, the production consequence, and the follow-up date. Five fields are enough to show whether the organization is learning how to choose under pressure or merely documenting that a conversation occurred.

This is where psychological safety supports physical safety. A worker who can raise a weak control without ridicule gives the system time to act before the exposure becomes an event. The article on three culture diagnostic lenses for plant leaders explains why survey results need worksite proof.

Leaders should also compare the decision log with the monthly dashboard. If the dashboard improves while concerns disappear, the organization may be seeing silence rather than safer work. If concerns increase and decisions become faster, the first signal may look worse before the system becomes more reliable.

5. The measured result was an outcome, not a method

The measured result was a 50% reduction in the accident ratio over 6 months, according to Andreza Araujo's documented PepsiCo South America experience. That number demonstrates what happened in one context, not what a generic program guarantees.

Separating outcome from method protects the case from becoming a slogan. The result should not be copied as a target without understanding the leadership choices, operating conditions, workforce, baseline, and controls that surrounded it.

The right question for a new operation is not whether it can reproduce 50%. The better question is whether it can show, within 6 months, that leaders protect critical controls more consistently, escalate conflicts earlier, and close decisions with accountable owners.

Use 4 measures together: accident outcomes, high-potential exposure, control verification, and decision follow-through. The first tells you what happened. The next 3 show whether the system is becoming more capable before the next event tests it.

6. What leaders had to stop doing

Leadership change becomes credible when managers stop rewarding the behavior that kept the old culture alive. A new message cannot coexist indefinitely with old consequences, because employees judge the response to pressure more heavily than the language of the policy.

Leaders must stop treating every deviation as an individual attitude problem, stop accepting a completed form as proof that a control worked, and stop asking EHS to own decisions that belong to operations. Those 3 habits make accountability look active while leaving risk ownership vague.

James Reason's work on latent failures helps explain why the visible action is not always the full cause. A supervisor may make the final choice, while staffing, planning, incentives, and weak escalation have already narrowed the available choices.

The corrective move is not to remove responsibility. It is to place responsibility at the right level. The person closest to the task should report the condition, the supervisor should control the immediate exposure, and the manager should remove the recurring system constraint.

7. The transferable lesson is a sequence

The transferable lesson is a sequence, not a branded intervention. Diagnose the gap, change the leader's response, make decisions traceable, verify critical controls, and only then interpret the trend in the accident data.

That sequence gives a 320-employee plant a practical starting point without pretending that every site has the same hazards or maturity. In the first 30 days, make behavior visible. By day 90, make decisions traceable. At month 6, compare the result with the evidence that preceded it.

Safety walks are useful when they test the real work rather than collect agreeable answers. The guide on 6 questions for safety walks can help supervisors turn field presence into evidence.

Andreza Araujo's experience across 30+ countries reinforces the same boundary. The language of safety changes by culture and industry, but employees everywhere notice whether leaders protect life when protection costs time.

Comparison between declared culture and operated culture

Declared culture describes what an organization says it values, while operated culture describes what people experience when a decision carries a cost. The comparison matters because the gap between those 2 conditions is where safety performance becomes fragile.

ISO 45001 specifies a management-system approach, while ILO guidance emphasizes that occupational safety and health depends on prevention, participation, and employer responsibility. Those authorities give the structure. Leadership behavior gives the structure operational meaning.

What to apply in your operation next

The next move is to choose one recurring pressure point and test the culture there for 12 weeks. A focused experiment is more useful than a broad campaign because it lets leaders see whether a changed response alters reporting, control verification, and decision speed.

Choose one work process, name one accountable leader, define 4 evidence fields, and review the record every week. At week 6, ask what became easier to report. At week 12, ask what decision changed because someone spoke early.

WHO describes mental and physical health as connected dimensions of well-being, and that principle also matters in safety leadership because fatigue, stress, and uncertainty can alter judgment before an incident appears. A culture that protects the person improves the quality of the decision.

Every week without a visible decision owner leaves the same conflict to be solved again at the point of pressure, where time is short and the cost of silence is highest.

Conclusion

The PepsiCo case shows that a 50% accident-ratio reduction in 6 months should be read as the visible result of pressure-tested leadership choices, not as proof that one campaign works everywhere.

Start with the next difficult decision. If leaders protect the control, give the concern an owner, and make the response visible, safety culture has moved before the metric. Explore Andreza Araujo's work on safety culture and leadership.