Safety Culture Transformation: What 250+ Projects Reveal

When a company says its safety culture has changed, the decisive evidence is not a completed training calendar. It is whether supervisors still protect a weak signal when production is under pressure. Across more than 250 companies served by ACS Global Ventures, Andreza Araujo's experience points to a practical conclusion: ownership appears only after leaders change the decisions that make risk visible or invisible.

1. Initial scenario: compliance was visible, ownership was not

A compliance-heavy operation can look disciplined while workers still learn that speed matters more than uncertainty. Procedures are signed, audits are closed, and safety messages are repeated, yet the shift team keeps unresolved hazards inside informal workarounds. That gap is the starting point for many culture transformations.

Andreza Araujo has worked across consumer goods, supply chain, mining, agriculture, construction, and industrial operations for more than 25 years, according to her professional profile. The recurring pattern is not a lack of rules. It is a mismatch between the rule leaders declare and the decision the operation rewards.

That mismatch is why a safety culture survey cannot stand alone. Perception data can show that trust is low, but field evidence is needed to show where the organization converts that perception into risk.

2. Decision: stop measuring culture by activity

The first decision is to stop treating activity as proof of transformation. Training hours, audit scores, and completed observations matter, although they do not demonstrate that a worker can challenge an unsafe plan without losing credibility or time.

In Safety Culture: From Theory to Practice, Andreza Araujo connects culture with observable habits and leadership choices rather than slogans. That distinction changes the diagnostic question from “How many initiatives did we launch?” to “Which decisions became safer because the initiative existed?”

A leadership team should therefore select a small set of decision points that reveal ownership. Examples include whether a supervisor pauses a job after a weak signal, whether a maintenance concern reaches the plant manager, and whether schedule recovery includes a documented risk review instead of a verbal assurance.

3. Execution: diagnose the work before prescribing the program

A transformation begins with evidence from the work, because a generic action plan usually reflects the language of the safety department rather than the operating system of the site. The diagnostic should follow how work is planned, authorized, performed, interrupted, and reviewed.

One useful sequence examines four layers. First, review the formal expectation. Next, observe the real task. Then compare the two with the worker and supervisor who carry the exposure. Finally, trace the decision rights that allowed the gap to persist.

This is where a safety culture evidence walk adds value. It does not replace a survey or audit. It connects what people report with what leaders can see, which makes the resulting plan specific enough to manage.

Across 30+ countries, Andreza Araujo has seen that the same word can hide different operating realities. “Empowerment” may mean stop-work authority in one plant and merely speaking during a toolbox talk in another. A credible diagnosis defines the behavior, the decision owner, the time window, and the evidence that would prove a change.

4. The operating change: make risk ownership concrete

Risk ownership becomes real when a named person must decide what happens next, with a clear escalation path and a time boundary. “Everyone owns safety” sounds inclusive, but it often leaves the hardest decision unowned when the line is behind schedule.

A practical plan assigns ownership at three levels. The frontline supervisor owns immediate control verification. The functional manager owns recurring conditions that cross shifts or departments. The site leader owns trade-offs that require capacity, maintenance, engineering, or production changes.

That design is consistent with the central argument of A Ilusão da Conformidade, which treats compliance as insufficient evidence that protection exists in practice. It also reflects James Reason's distinction between active failures and latent conditions, because a visible mistake may be the final expression of a decision made much earlier.

5. What changed across the portfolio

The measurable lesson from a portfolio of projects is not that every site achieves the same result. It is that transformation becomes more reliable when the organization measures a chain of cause and effect instead of one headline rate.

Andreza's documented experience includes a 50 percent reduction in accident ratio over six months under a 180-day plan at PepsiCo South America Foods. That result is a verified case, not a promise that every program will reproduce the same number. Its strategic value is the sequence behind the result: leadership attention, disciplined execution, and a decision system that made risk harder to ignore.

A site can track whether the chain is operating by reviewing the age of unresolved high-risk actions, the percentage of critical controls verified by the accountable owner, the time between a reported weak signal and a management decision, and the proportion of recurring findings that return after closure. These measures are useful because they expose movement before an injury rate changes.

6. The leadership routine that protects the result

Culture change fades when the leadership routine stops asking for evidence. A monthly steering meeting is not enough if supervisors receive no response during the shift and site leaders only review the dashboard after the exposure has passed.

The routine should connect four moments. At the start of work, supervisors identify the conditions that could change the plan. During execution, they verify the controls that matter for serious harm. After a deviation, the accountable manager decides whether the work can continue. At the weekly review, leaders remove the recurring barrier rather than simply recording the event.

The experience of aligning safety decisions across countries shows why cadence matters in distributed operations. A common principle is useful, but each site still needs a local mechanism for escalation, response, and proof that a decision was implemented.

In Make The Difference: Be a Leader in Health and Safety, Andreza Araujo frames leadership as an operating behavior. That framing is important because presence without follow-through can create visibility while leaving the underlying risk unchanged.

7. Measured result: separate transfer from attribution

A responsible case study distinguishes what was measured from what was inferred. If a transformation coincides with a lower accident ratio, the article should identify the verified result and explain the management changes that plausibly contributed without claiming that one intervention caused every outcome.

This discipline protects credibility. It also helps executives decide what to repeat. The transferable elements are usually the decision architecture, the review cadence, and the ownership model. The local numbers depend on exposure, workforce stability, contractor mix, reporting behavior, and the quality of the controls already in place.

That is why leaders should publish a before-and-after view with both outcome measures and mechanism measures. A lower rate belongs beside control-verification quality, action aging, escalation time, and recurring-condition closure. Without the mechanism measures, a favorable result can hide a reporting change or a temporary reduction in exposure.

8. Generalizable lessons for the next transformation

The first lesson is that culture cannot be delegated to communication. Messages support change, but operating decisions determine whether the message survives contact with a difficult shift.

The second lesson is that diagnosis must reach the point where risk is accepted, deferred, or escalated. Surveys and audits become more useful when they identify the decision that produced the observed condition.

The third lesson is that a portfolio result should guide questions, not justify a promise. Andreza Araujo's 25+ years, 250+ companies, and 30+ countries provide an experience base for pattern recognition, while each site still needs its own evidence.

The fourth lesson is that ownership needs a visible owner, a time boundary, and a verification method. Without those three elements, “accountability” becomes a meeting phrase.

Conclusion: ownership is the result leaders can observe

Across more than 250 companies, the durable lesson is clear. Safety culture changes when leaders redesign the decisions, escalation paths, and verification routines that shape work under pressure. Training and campaigns can support that shift, but they cannot substitute for it.

If your operation needs to move from visible compliance to credible ownership, begin with one high-risk decision, name its owner, define the evidence, and review the result within a fixed cadence. Safety is about coming home, and the culture that supports that outcome must be visible in the work itself.