The PepsiCo Supervisor Test: What a 50% Accident-Ratio Reduction Reveals About Safe Behavior

A safety result can look simple after it has been reported. PepsiCo South America reduced its accident ratio by 50% in six months, according to the case record described in Andreza Araujo's executive work. The number is memorable, but it does not explain what changed at the point of work.

The deeper question is harder and more useful. What did supervisors do differently when a task moved away from the plan, a control became inconvenient, or production pressure made the safer choice slower? That question turns the case from a success headline into a test for safe behavior.

What did the PepsiCo result actually measure?

The case covered a large South America operating context that included seven countries, 30 factories, and 168 distribution centers. The reported result was a 50% reduction in accident ratio over six months. Those figures describe the scale and the outcome, but they do not prove that one intervention caused every improvement.

That distinction matters because accident ratios are lagging measures. They tell leaders that the frequency of recorded harm changed, yet they do not show whether a worker could stop a task, whether a guard was usable, or whether a supervisor responded to a weak signal before harm occurred. James Reason's work on latent failures helps explain why a visible result can sit on top of less visible changes in decisions, design, workload, and supervision.

The right reading is therefore not that a company found a formula that every site can copy. The right reading is that a large network created enough alignment for local leaders to make safer work more executable. Safe behavior was the observable outcome of that operating discipline, not a motivational message added after the fact.

Initial scenario

A multi-country operation has a familiar problem. Corporate leaders publish a standard, the EHS team trains supervisors, and every site reports activity. The documents look aligned, although the decisions made during a rushed changeover, a maintenance delay, or an unusual delivery may still vary from one location to another.

In that environment, workers often receive a safety instruction without receiving the conditions that make it practical. A supervisor may know that a task requires isolation, a second check, or a pause, yet still lack the time, authority, equipment, or escalation route to protect that decision when the schedule starts to move.

The danger is not limited to intentional shortcuts. A person can choose a reasonable action inside a constrained system and still create exposure because the available options are weak. A useful safe-behavior program must examine that choice without losing sight of the control that should have supported it.

This is why the PepsiCo result should be read at supervisor level. Supervisors sit close enough to see the task and close enough to influence the next decision. They are often the first people who can make a control credible, or make it disappear through an exception that nobody records.

The decision supervisors had to make

The central supervisor decision was not whether safety mattered. Most organizations already answer that question correctly. The practical decision was whether the safer method would remain the normal method when it took more time, required a production conversation, or exposed a weakness in the plan.

A supervisor who handles that moment well does four things. First, the supervisor identifies the actual task rather than relying on the planned description. Second, the supervisor checks whether the critical control is present and usable. Third, the supervisor clarifies who has authority to pause or change the work. Fourth, the supervisor makes the response visible so that the next shift does not repeat the same uncertainty.

Those actions turn a broad value into a behavior that a worker can observe. They also create evidence that leaders can review. A conversation becomes more than a conversation when it changes the work method, resolves an interface, protects a stop decision, or assigns an owner with a clear verification point.

Andreza Araujo's book Safety Culture: From Theory to Practice is useful here because culture is treated as something visible in decisions and routines, not as a sentence on a wall. The supervisor test asks whether the stated standard survives contact with the actual task.

Execution at the workface

The first execution move is to define the behavior in operational terms. “Work safely” is too broad for a supervisor to verify. “Confirm the isolation boundary before opening the line, and stop when the boundary does not match the permit” creates a decision that can be seen, discussed, and checked.

The second move is to observe the condition around the behavior. A worker who bypasses a step may be responding to a missing tool, an unclear permit, a late handover, or a control that cannot be reached without creating another exposure. A supervisor who records only the action will miss the reason the action made sense at that moment.

The third move is to close the response loop. The supervisor should name the change, identify who owns it, and return to the task to verify that the change is still present. This is where many programs lose credibility. The observation is logged, the conversation is praised, and the same work condition remains in place.

The fourth move is to make escalation normal. When the local team cannot fix the condition, the supervisor needs a route that preserves the issue rather than rewarding silence. Escalation is a safe-behavior control because it protects the person who reports the mismatch and gives the organization a chance to remove the barrier.

The distinction between attendance and demonstrated competence provides a practical companion to this case. A worker can complete a module and still be unable to explain what changes when the task changes. A field verification should therefore test the decision, the control, and the response to uncertainty, not only the worker's memory of the training.

Measured result

The reported 50% reduction in accident ratio over six months is the measured result attached to the PepsiCo South America case. It is significant because the operating context was large and distributed, with seven countries, 30 factories, and 168 distribution centers. It is also limited because the ratio does not identify the contribution of every local decision.

The table is not a reconstruction of a PepsiCo dashboard. It is a practical interpretation of what a credible safe-behavior system must make visible if a reported result is to mean more than a change in the lagging number.

Generalizable lessons

First, behavior follows the available decision. A worker cannot consistently choose a control that the system makes unavailable, unclear, or unsafe to use. Leaders should ask what the worker was permitted to do, not only what the worker did.

Second, supervisor quality is a control. A supervisor can strengthen a barrier by checking it, protecting time for it, and escalating when it fails. The same supervisor can weaken it by accepting an undocumented exception. This is why observation volume is a weak substitute for response quality.

Third, a result needs a mechanism. The 50% figure matters because it points to a transformation that must be explained through operating choices. Without a mechanism, the number becomes a story that other companies imitate without understanding what made it possible.

Fourth, local transfer matters more than central language. A network can share one expectation while allowing each site to express the decision in the language of its work. Consistency means preserving the control intent, not forcing every conversation into the same script.

Fifth, psychological safety supports the data. Workers need enough trust to say that a procedure cannot be followed, a control is missing, or a plan has changed. Without that voice, leaders may see compliance while the task is already operating outside the intended protection.

What to apply in your operation

A plant manager can run the supervisor test in one week without launching a new campaign. Select one task with meaningful exposure and ask five questions during the next field review.

• What decision must the supervisor make before the task starts?
• Which control should make that decision easier?
• What condition would require a pause or escalation?
• Who owns the response when the control is missing?
• When will someone return to verify that the change worked?

Use the answers to build a short field check, then compare the written plan with the work as performed. If the gap is caused by equipment, redesign the control. If it is caused by authority, clarify the decision right. If it is caused by workload or timing, change the plan instead of asking the worker to absorb the conflict.

Leaders should also review the age and quality of responses. A closed action is not necessarily a corrected exposure. The stronger indicator is whether the action changed the task and whether the people who perform the task can explain the new decision path.

The article on leadership quality in behavior-based safety projects adds a useful test for response quality. The guide on training attendance versus demonstrated competence helps separate a record of participation from evidence that behavior transferred to the field. For procedure-heavy work, four procedure-transfer gaps show where written intent can disappear before the task begins.

The case lesson for safe behavior

The PepsiCo South America result should not be copied as a slogan. Its value is diagnostic. A 50% accident-ratio reduction in six months is most useful when leaders ask which supervisor decisions, control changes, and escalation habits made safer work more executable across a distributed operation.

That is the supervisor test. If the organization wants safer behavior, it must make the safer decision workable, visible, and protected when conditions change. The number belongs in the history of the case. The operating discipline belongs in the next task.

Explore the Headline Podcast for conversations about the leadership decisions that shape safety culture and better workplaces.