How a consumer-goods leader used 3 moves to turn a 50% accident reduction into culture change in 6 months

During her tenure at PepsiCo South America, where the accident ratio fell 50% in six months, Andreza Araujo learned a lesson that many safety programs still miss. A fast result is not the same as a durable result. If leaders do not change the routines that create risk, the next pressure cycle will try to rebuild the old pattern.

A real safety turnaround survives only when leaders change three things at the same time: who can decide, what the daily review inspects, and how the field confirms that controls are still real.

That is the thesis of this case study. As Andreza Araujo argues in Safety Culture: From Theory to Practice, culture is visible in repeated decisions, not in slogans. In more than 250 cultural transformation projects across 30+ countries, the same pattern appears again and again: when leaders fix the operating rhythm, the field changes faster than the slide deck.

Initial scenario

The starting point was not indifference. It was an operation that still leaned on reaction. Leaders cared, supervisors tried hard, and the site could produce reports, but the management system still spent too much time explaining the last event and too little time shaping the next one.

Across 25+ years leading EHS at multinationals, Andreza Araujo has seen that many plants ask for more training when the actual problem sits in decision flow, supervision, and the way weak signals are handled. James Reason is useful here because he shows how latent failures line up quietly before they surface as harm. Patrick Hudson is useful too, because his maturity model explains how an organization can look structured while still depending on a few strong people to keep things from drifting.

If you want the evidence side of that problem, the article on safety culture traces shows the kinds of field signals that matter, while cosmetic compliance shows why a clean dashboard can still hide a weak operation.

Decision

The leadership decision was to stop treating accident reduction as a campaign and start treating it as a management design problem. That shift matters because campaigns are temporary, while routines are what people actually repeat when pressure rises.

The first move was to make decision rights explicit. Someone had to know when work could continue, when it had to stop, and when an exception had to move beyond the shift. The second move was to make field verification part of leadership work, not an extra chore for the safety team. The third move was to make supervisors own the follow-through on weak signals, so the organization did not confuse presence with control.

This is where visible felt leadership stops being a phrase and becomes a practice. Presence only matters when it changes what a supervisor asks, checks, and closes. If the manager walks the floor but leaves every decision untouched, the site has movement, not leadership.

Andreza Araujo often frames this point in Safety Culture Diagnosis: Learn how to do your own. A site does not become safer because leaders say they care. It becomes safer when the system makes the next decision easier than the old one.

Execution

The execution phase was practical and unglamorous. The site moved from broad messages to specific routines. Daily reviews stopped focusing only on status and started focusing on open controls, delays in escalation, and places where a supervisor was carrying work that should have been visible to the line. That change reduced the space where drift could hide.

One useful lesson from this case is that training did not lead the change. Training helped only after the work itself changed. That is a key trap the market minimizes. If the job still rewards speed over verification, no amount of classroom time will keep the field honest for long. The supervisor will still feel pressure to get through the shift, and the crew will still look for the shortest path around the slowest barrier.

As Andreza Araujo writes in A Ilusão da Conformidade, translated as The Illusion of Compliance, a clean document can hide a dirty process. The point is not to reject procedure. The point is to make the procedure match the work people actually do.

For that reason, the article on building a safety decision trail is a useful companion. If a leader cannot show what changed, who decided, and when the control was verified, the improvement story is too fragile.

Measured result

The headline result was a 50% reduction in accident ratio in six months. That number matters, but only because it was tied to a visible change in how the site made decisions. Leaders stopped waiting for the monthly report to tell them what the field already knew. Supervisors stopped carrying weak signals alone. The line no longer depended on a few heroic people to keep the barriers upright.

That pattern is why a site can improve fast without becoming superficial. The reduction is real, yet the deeper result is that people now expect managers to look at work, not only at reports. That expectation is culture change.

Generalizable lessons

Three lessons matter most.

• Culture changes when decision rights are explicit. If nobody knows when to stop work, the lowest-friction choice becomes the default.
• A metric without field verification is easy to flatter. The number may move while the same bad habits stay alive underneath it.
• Training alone is too weak when the job still rewards speed over control. The work has to change, because people adapt to the system they inherit.
• Managers should inspect the barrier, not the story. If the control is failing in the field, a nicer narrative will not fix it.

Those lessons fit James Reason's view of latent failure, and they fit Andreza Araujo's own view in Safety Culture: From Theory to Practice. Safety turns durable when leaders stop asking the field to compensate for vague authority, weak follow-up, and invisible drift.

If you want a practical counterpart to this lesson set, management review traps shows how leadership meetings become ceremony, while culture traces shows how to test whether the system really changed.

What to apply in your operation

If you want to use this case in a plant, warehouse, or multi-site operation, start with one routine that repeats every week and one risk that people already know but keep normalizing. Do not begin with a broad safety campaign. Begin with the one decision that controls whether the work continues, pauses, or escalates.

Then do four things. Name the decision owner. Put one field verification into the daily review. Write the open control, not just the action item. Close the week with a note about what changed in work, not only what was discussed in a meeting.

This is where the gap between a good metric and a good system becomes obvious. A manager can report progress and still leave the old habit untouched. A better manager makes the new habit easier to repeat, which is why the next shift does not have to rediscover the same risk.

For leaders who want to go deeper, the books and programs by Andreza Araujo give more than a slogan. They show how to read culture, diagnose drift, and turn a metric into a management decision.

FAQ

Why does a 50% reduction not automatically mean culture changed?
Because a metric can improve while the same habits remain in place. Culture changes only when leaders change decision rights, review routines, and field verification. Andreza Araujo treats the metric as evidence, not as the whole story.

What changed first in this case?
The first change was in the operating rhythm. Leaders made decisions more explicit, moved field verification into leadership work, and used daily reviews to surface weak signals before they turned into repeat risk.

How should a plant manager use this lesson?
Pick one recurring risk, define who can stop or re-sequence the work, and add one verification step to the daily review. If the manager cannot show what changed in the work, the improvement is still too fragile.

Why did training not lead the change?
Because training only works when the job itself is designed to support it. If the work rewards speed over verification, people will adapt to the pressure and look for shortcuts unless the operating system changes too.

Which Andreza Araujo book fits this topic first?
Safety Culture: From Theory to Practice is the best starting point because it explains why culture shows up in repeated decisions. Safety Culture Diagnosis: Learn how to do your own is the next step if you want to turn that idea into evidence.

The real test is simple. If the next pressure cycle arrives, does the site still depend on heroics, or does it have a routine that holds without noise? If you need help checking that difference, request a safety culture diagnostic or start with Safety Culture: From Theory to Practice and Safety Culture Diagnosis: Learn how to do your own.