Safety Culture Diagnosis: 250-Company Case

Safety culture diagnosis fails when leaders treat it as a survey event instead of a management decision. The useful case is not the company that collected thousands of answers. The useful case is the organization that converted weak signals into a roadmap, changed leadership routines, and returned evidence to the field within the first 90 days.

This case study draws from the pattern Andreza Araújo has seen across 250+ companies and 30+ countries, including multinational operations in consumer goods, supply chain, mining, agriculture, construction, and industrial services. The central lesson is uncomfortable for senior leaders: culture does not improve because a dashboard says engagement is high. It improves when the diagnosis exposes where declared values fail under production pressure.

As Andreza Araújo argues in Safety Culture Diagnosis: Learn how to do your own, the diagnostic process must reveal the gap between what people say in official forums and what they do when a supervisor, contractor, or operator has to choose between speed and control. A score without that gap is only reputation management.

Initial scenario

A safety culture diagnosis normally begins with a contradiction: executives believe the system is mature because ISO 45001 routines, training records, committees, and incident reports exist, while the field experiences those routines as paperwork that rarely changes risk. In the 250+ company pattern observed by Andreza Araújo, the first problem is not a lack of safety language. It is the excess of safety language without proof that people can challenge unsafe work.

The typical starting point is a site with visible programs and weak trust. The company has a monthly safety meeting, behavioral observations, a risk register, a training matrix, and a reporting channel. It also has contractors who do not stop work, supervisors who solve deviations quietly, and workers who know which hazards are safe to mention. That split is why a diagnosis has to combine survey data, interviews, field observation, document review, and leadership behavior, not only a perception form.

ISO 45001 specifies worker consultation and participation as part of the occupational health and safety management system, and that requirement matters because a culture diagnosis is a test of participation in real conditions. If workers report hazards but never hear what changed, participation is administrative. If supervisors listen only after an injury, participation is late. If managers reward the team that finishes the job fastest regardless of the controls bypassed, participation is cosmetic.

The initial scenario is therefore not a broken culture in the abstract. It is a management system with a weak feedback loop. This is where the diagnostic work must begin.

Decision

The decisive move in a safety culture diagnosis is to stop asking whether people like the safety program and start asking whether the program changes decisions under pressure. Across 25+ years leading EHS at multinationals, Andreza Araújo has observed that mature cultures are visible in small decisions: who can interrupt a task, who owns a corrective action, who returns a lesson to the crew, and who loses status when production and protection collide.

That decision changes the diagnostic design. Instead of a generic climate survey with pleasant scores, the company needs a diagnostic question set tied to control ownership, leadership response, reporting closure, contractor voice, and serious risk exposure. The useful survey is not the one that flatters leadership. It is the one that shows whether a permit-to-work is read, whether a safety climate survey is hiding fear, and whether a near-miss report produces action or silence.

In Safety Culture: From Theory to Practice, Andreza Araújo connects maturity with visible leadership behavior rather than declared intent. That distinction is important because leaders often want culture to be measured as sentiment. Real diagnosis measures whether the organization can detect, discuss, and correct risk before an event forces the conversation.

The practical decision is to treat diagnosis as the first step of a 90-day intervention. The output should not be a PDF with colorful charts. It should be a ranked roadmap with owners, deadlines, and evidence requirements for each cultural gap.

Execution

A credible safety culture diagnosis uses at least 5 evidence streams: perception survey, leadership interviews, field observation, document sampling, and incident or near-miss pattern review. The 250+ company pattern shows that each stream corrects the others, because surveys can be polite, interviews can be strategic, field observations can be staged, documents can be perfect, and incident data can be underreported.

The first execution step is segmentation. A single culture score hides the operating truth. A plant manager may see one number, although maintenance, logistics, contractors, night shift, and supervisors are living different cultures in the same facility. The diagnosis should separate at least role, shift, contractor status, location, tenure, and exposure type, because a 4.2 average can hide one crew where voice is strong and another where people have stopped reporting.

The second step is triangulation. If workers say stop-work authority is trusted but the last 12 high-risk deviations were resolved without a pause, the system is not as trusted as the survey suggests. If supervisors say corrective actions are closed quickly but workers cannot name a single change returned from the last report, closure is not experienced by the people who raised the risk. That is why the diagnostic team must compare perception with evidence.

The third step is leader calibration. Senior managers need to see the same field evidence that workers see, not only the summarized score. During the PepsiCo South America tenure, where the accident ratio fell 50% in 6 months, Andreza Araújo learned that performance shifted when leaders disciplined their routines around visible risk review, fast response, and credible follow-up. The same principle applies here. Diagnosis becomes action only when leaders change what they inspect, ask, reward, and tolerate.

Measured result

The measured result of a safety culture diagnosis is not the score itself. The measured result is the quality of the decisions made after the score appears. In the cases Andreza Araújo has supported, the strongest early indicator is whether leaders can name the top 3 cultural gaps, assign owners, and show field evidence that something changed within 30, 60, and 90 days.

A useful diagnosis produces several measurable outputs. It identifies the weakest groups by role or location, ranks cultural gaps by risk relevance, separates communication problems from control problems, and shows whether the organization is closer to reactive, calculative, proactive, or generative maturity. It should also tell leaders which rituals are consuming time without reducing exposure.

The first 30 days should focus on credibility repair. Workers need to see that the diagnosis was not extracted from them and then hidden from them. The company should return the findings to supervisors and crews in plain language, select 3 priority gaps, and communicate what will change. A quiet diagnosis is worse than no diagnosis because it proves the same cultural weakness the survey was supposed to reveal. The 19-country safety culture scale-up case shows why this feedback rhythm matters when the organization spans several operating realities.

By 90 days, the company should have evidence in 4 places: changed leadership routines, closed feedback loops, corrected high-risk rituals, and better quality of safety conversations. If the only measurable result is a prettier maturity chart, the diagnosis has remained cosmetic. If corrective action closure improves, stop-work pauses become less rare, and near-miss quality rises, the diagnostic process has started to alter the culture.

Generalizable lessons

The first generalizable lesson is that diagnosis must be specific enough to threaten comfortable narratives. A company that only asks whether people value safety will receive a polite answer. A company that asks whether workers can challenge a supervisor, whether contractors receive the same protection as employees, and whether a report gets answered within 48 hours will see the culture more clearly.

The second lesson is that maturity models help only when they change management behavior. The Bradley, Hudson, and Hearts and Minds comparison is useful when leaders use it to decide what to change next. It becomes theater when it is treated as a label. Calling a culture proactive does not make it proactive.

The third lesson is that weak feedback loops destroy trust faster than weak speeches can repair it. Andreza Araújo often frames culture as what the organization repeatedly teaches people to expect. When a worker reports a hazard and hears nothing for 3 weeks, the culture has taught a lesson. When a supervisor pauses production to verify a control and receives support, the culture has taught a different one.

The fourth lesson is that the diagnosis must reach contractors. Many serious exposures sit at interfaces: maintenance, logistics, construction, cleaning, temporary labor, and simultaneous operations. A diagnosis that measures employees while ignoring contractors can make the core organization look mature while the highest-risk work remains culturally invisible.

What to apply in your operation

Safety leaders can apply the 250+ company pattern by converting diagnosis into a short, disciplined operating cycle. The cycle starts with evidence collection, moves into prioritization, assigns owners, returns findings to the field, and then measures whether leadership behavior changed within 90 days. The cadence matters because culture decays when feedback is slow.

Start by choosing a diagnostic scope that matches real risk. A corporate survey across 20 countries may be useful, but a high-risk plant with weak contractor voice may need a deeper local diagnosis before the global score means anything. Then define the evidence streams before the survey launches. Decide what documents will be sampled, which crews will be observed, which leaders will be interviewed, and what incident patterns will be reviewed.

After the results, resist the temptation to fix everything. Select the few gaps that connect most directly to SIF exposure, reporting quality, control verification, or leadership credibility. If the diagnosis shows that workers do not trust corrective action closure, do not launch a poster campaign about care. Fix the closure loop and return proof to the people who reported the risk.

For organizations ready to move beyond measurement, Safety Culture Diagnosis: Learn how to do your own gives practitioners a structure for turning perception into action. The field test is simple: 90 days after the diagnosis, workers should be able to name what changed because they spoke honestly.

FAQ

What is a safety culture diagnosis? A safety culture diagnosis is a structured evaluation of how people perceive, discuss, and manage risk in daily work. It usually combines survey data, interviews, field observation, document review, and incident pattern analysis. The goal is not to produce a score, but to identify where leadership routines, feedback loops, and control ownership are failing.

How long should a safety culture diagnosis take? A focused site diagnosis can often be completed in 30 to 60 days, depending on workforce size, contractor exposure, shifts, and geography. Multi-country assessments may require longer, but the first field feedback should not wait for a final corporate report. A 90-day action cycle keeps the diagnosis connected to visible change.

What should leaders measure after the diagnosis? Leaders should measure whether priority gaps were assigned, whether feedback was returned to crews, whether corrective action closure improved, whether near-miss quality changed, and whether supervisors altered daily routines. A maturity score without these follow-up measures is weak evidence. Andreza Araújo's diagnostic approach treats the score as the beginning of management work.

How is safety culture diagnosis different from a safety climate survey? A safety climate survey captures perceptions at a point in time, while a safety culture diagnosis compares those perceptions with field behavior, documents, leadership interviews, and incident patterns. The difference matters because a survey can be positive even when controls are weak. This distinction is expanded in the article on safety climate survey blind spots.

Which maturity model should a company use? Bradley, Hudson, and Hearts and Minds can all support a diagnosis when the organization uses the model to guide action rather than decoration. The best model is the one leaders can translate into field routines, feedback loops, and risk decisions. The comparison is covered in Bradley vs Hudson vs Hearts and Minds.

Final field test

A safety culture diagnosis earns its place only when people can see what changed because they told the truth. The evidence is not the chart, the score, or the board slide. The evidence is the supervisor who responds differently, the contractor who can pause unsafe work, the report that receives a timely answer, and the leadership meeting that now starts with risk evidence instead of only lagging indicators.

Andreza Araújo's work across 250+ companies shows that diagnosis is powerful when it becomes a management operating rhythm. Safety is about coming home, and that requires more than asking people what they think. It requires proving, within 90 days, that honest answers can change the way the organization leads.

To apply this approach with your leadership team, start with Andreza Araújo's safety culture transformation work or deepen the method through the book Safety Culture Diagnosis: Learn how to do your own.