How PepsiCo South America Cut Accident Ratio by 50% in Six Months

Initial scenario: a strong result still needs a careful reading

A 50% reduction in accident ratio over six months attracts attention because it compresses a difficult safety problem into one visible number. Andreza Araujo’s documented PepsiCo South America result is precisely that kind of case. During her tenure as EHS and Sustainability Manager for PepsiCo Foods across Latin America and the Caribbean, the business operated across seven countries, 30 factories, and 168 distribution centers, and a 180-day plan produced a 50% reduction in accident ratio in six months.

That result matters, yet the number is not the whole case. An accident ratio is a lagging indicator. It tells leaders what happened after exposure, control failure, and harm have already passed through the system. It does not identify which decisions changed first, whether risk was redistributed, or whether reporting behavior changed at the same time.

That distinction protects the case from becoming a success story with no usable method. The documented result belongs to a real operating context. The transferable lesson must therefore be framed as a leadership discipline, not as a promise that every plant can reproduce 50% in the same period.

Decision: treat the metric as a signal, not as the strategy

The first decision for a leadership team is to refuse the easiest interpretation. If the only question is how to reduce the accident ratio, the organization may respond with a campaign, a training calendar, or a request for more observations. Those actions can be useful, although none of them explains which exposures deserve executive attention.

ISO 45001:2018 requires organizations to understand hazards, risks, opportunities, operational controls, worker participation, and improvement. The standard does not turn one outcome metric into a complete picture of safety performance. OSHA recordkeeping guidance serves a different purpose as well. It creates a formal record of work-related injuries and illnesses, but a record is not the same thing as a live view of critical risk.

The leadership decision should therefore be phrased more precisely. What must change in the next 180 days so that serious exposure is detected, owned, controlled, and verified before it becomes another recordable event? That question keeps the metric in its proper place while directing management attention toward the conditions that precede it.

Execution: what the documented case allows us to say

The public case record gives us a measured outcome and a time window. It does not provide a license to invent internal meeting agendas, plant-by-plant interventions, or a universal sequence of actions. A responsible article must preserve that boundary, because a fabricated detail can make a case sound more practical while making it less trustworthy.

What can be inferred is the management architecture that any comparable turnaround would need. A 180-day plan requires a defined owner, a short decision cycle, a method for prioritizing exposure, and a review rhythm that connects field evidence to executive action. Without those elements, the plan becomes a calendar rather than a control system.

Across 25+ years leading EHS at multinationals, Andreza Araujo has consistently positioned safety as an operating responsibility rather than an isolated specialist activity. Her book Safety Culture: From Theory to Practice makes the same distinction in cultural terms, because a stated value only becomes credible when routines, choices, and consequences reinforce it.

The strongest reading of the PepsiCo result is therefore not that one intervention caused a percentage to fall. It is that a defined leadership plan created enough focus for multiple controls and decisions to move in the same direction, which is the condition a different organization must test for itself.

Measured result: what changed, and what the number cannot prove

The table shows why precision matters. The result is strong enough to anchor a case study, but it cannot establish causality for every action that a reader might want to copy. It also cannot tell a board whether the organization became safer against serious injury and fatality exposure, because ordinary accident ratios can be dominated by lower-severity events.

That is why the International Labour Organization’s safety and health guidance remains relevant. Prevention depends on safe systems of work, worker participation, competent supervision, and conditions that reduce exposure, not only on counting outcomes after the fact. The accident ratio should open the review, not close it.

Generalizable lesson 1: define the exposure before defining the target

A target without an exposure map invites activity that looks productive but does not reach the highest consequence risks. The leadership team should identify where energy, movement, height, pressure, chemicals, vehicles, and production interfaces can produce serious harm, then assign a control owner for each material exposure.

James Reason’s work on organizational accidents is useful here because it directs attention toward latent conditions and multiple defensive layers. A supervisor may be the last person seen near an event, although the opportunity to prevent it may have existed earlier in design, planning, maintenance, procurement, or staffing.

Andreza Araujo’s Safety Culture Diagnosis offers a complementary discipline. Diagnosis should reveal how people perceive the formal system, where the operating system diverges from it, and which leadership decisions keep that divergence in place. The purpose is not to produce another maturity label. It is to decide what deserves intervention first.

Generalizable lesson 2: pair lagging results with leading evidence

A six-month turnaround needs a dashboard that can show movement before the accident ratio changes. Useful evidence may include the percentage of critical controls verified in the field, the time between a reported concern and an assigned owner, the age of overdue corrective actions, the quality of worker participation, and the rate at which repeated hazards are eliminated rather than re-described. Leaders can compare this logic with the leading-indicator approach used for work-design risk.

These indicators do not become reliable merely because they are leading indicators. A completed observation can be a formality, and a closed action can be a paperwork event. Leaders must sample the evidence, ask what changed in the task, and check whether the same exposure returns under a different name. The same test applies to a corrective-action verification review.

The National Safety Council and regulator guidance across jurisdictions distinguish between recording performance and preventing harm, even though organizations often collapse both functions into one scorecard. The practical test is simple. Can the leadership team use the indicator to make a different decision this week?

Generalizable lesson 3: make worker voice part of verification

Worker participation is not an ornament added to a management system after the controls are designed. People who perform the task can identify workarounds, conflicting priorities, weak isolation points, and production pressures that a document review will miss. Their contribution becomes meaningful when the organization shows what it changed because someone raised a concern.

Psychological safety does not mean removing competence requirements or accountability. Amy Edmondson’s research supports a more disciplined interpretation, in which people can question a decision, report a problem, and ask for help without humiliation or retaliation. That condition improves the quality of information available to the people who own risk.

For the PepsiCo case, the transferable question is not how many conversations occurred. It is whether the 180-day plan made it easier for relevant information to reach the decision-maker while the exposure was still controllable.

What leaders should apply in the next 180 days

A company that wants to learn from the case should begin with a bounded review rather than a slogan. The executive sponsor can select the highest-consequence exposures, name the operational owners, and define the evidence that will show whether controls work under real conditions.

• Days 1 to 30: confirm the baseline, identify critical exposures, test the quality of current incident and near-miss data, and assign decision rights.
• Days 31 to 90: verify controls in the field, remove repeated barriers, review worker concerns with named owners, and escalate decisions that remain unresolved.
• Days 91 to 180: compare outcome movement with leading evidence, test whether improvements survive operational pressure, and decide which routines belong in the normal management system.

The sequence should remain flexible because the starting conditions will differ. A distribution center with vehicle interaction risks does not need the same review as a food plant with machine guarding and contractor interfaces. The discipline is consistent, while the control set must fit the exposure. That is also why leaders should define risk decision boundaries before the plan starts.

The case’s lasting lesson is disciplined leadership

Andreza Araujo’s PepsiCo result deserves attention because a 50% accident-ratio reduction in six months is a meaningful documented outcome. It should not be repackaged as a universal formula, and it should not be weakened into a motivational anecdote.

The useful lesson is more demanding. Leaders must connect a time-bound plan to exposure, ownership, worker voice, control verification, and evidence that survives production pressure. When those decisions become visible, the accident ratio can improve as a consequence of better risk management rather than as the sole object of management attention.

Safety is about coming home. For a leadership team, that means treating the number as an invitation to inspect the system that produced it.