Multi-Site Culture Baseline: 250+ Company Case

Multi-site culture baseline is the disciplined diagnosis of how different locations actually make safety decisions, not only how they describe safety in policies or surveys. It compares field evidence, leadership routines, supervisor behavior and worker voice so executives can see where culture supports risk control and where it only performs compliance.

A multi-site safety culture program usually starts with a comforting assumption, because leaders believe that one corporate standard produces one operating culture. The case pattern Andreza Araujo has seen across 250+ companies and 30+ countries says otherwise. Plants under the same brand, with the same policy and the same audit protocol, often run different cultures by Monday morning.

The useful lesson is not that standardization is weak. ISO 45001:2018 specifies leadership, worker participation and operational planning requirements precisely because the management system has to be lived through decisions. The gap appears when a company measures culture through opinion alone, then wonders why the next serious exposure comes from a site with good scores.

Initial scenario

The initial scenario in multi-site culture work is almost always a false average that hides local risk. A global or regional EHS leader receives a single culture score, a dashboard of training completion and a calendar of safety walks, although each site may have a different relationship with production pressure, stop-work authority and supervisor ownership.

HSE in the United Kingdom reports safety statistics by sector because the exposure profile of each operation matters, and the same logic applies inside a company. A warehouse, a process plant, a fleet route and a maintenance shutdown do not create the same decision environment. When the baseline ignores those differences, it rewards the site that fills records cleanly, not necessarily the site that controls risk.

Andreza Araujo argues in Safety Culture Diagnosis that culture diagnosis has to convert perception into observable evidence. That distinction matters for EHS managers who already have survey data but still cannot explain why a strong score coexists with weak escalation, repeated shortcuts or supervisors who tolerate small deviations until they become routine.

This is why a baseline should connect with existing diagnostic work instead of replacing it. A site that performs frequent walks can still miss the evidence problems described in safety walks that keep culture cosmetic, especially when leaders count visits but do not test whether controls changed after the visit.

What changed when opinion stopped being enough?

Opinion stopped being enough when leaders treated the baseline as a decision instrument rather than a climate survey. The change was small in language but large in consequence, because each site had to show evidence of ownership, escalation, correction and learning before its culture score could be trusted.

Across 25+ years in multinational EHS leadership, Andreza Araujo has identified a recurring trap. The safer site is not always the site with the most enthusiastic language about safety. It is the site where a supervisor can stop work without social punishment, where a maintenance manager restores a control before restarting, and where the plant manager asks about weak signals before reviewing the injury rate.

James Reason's work on latent failures helps explain the pattern. Serious events rarely come only from the last person touching the job. They develop through conditions that leadership has normalized, such as unclear authority, poor design, weak follow-up and controls that no one verifies after the audit closes.

The decision: build a baseline from field evidence

The central decision was to build the baseline from field evidence, not from declarations. That meant treating interviews, observations, document checks, supervisor routines and leadership decisions as separate lenses whose contradictions were as important as their agreements.

A practical multi-site baseline usually needs at least 4 evidence streams. The first is leadership routine, which shows what managers ask about every week. The second is supervisor behavior, which shows what gets tolerated during work. The third is worker voice, which shows whether reporting feels useful or risky. The fourth is control evidence, which shows whether critical risks are actually governed.

That structure also differentiates a baseline from a one-time audit. An audit can verify whether a procedure exists, but the baseline asks whether people use the procedure when production is late, when a contractor pushes back, or when the easiest path is to accept a temporary deviation. This is where culture becomes visible.

For new leaders, the same logic appears in a 30-day safety culture onboarding audit, because the first month is often when the leader discovers whether declared standards match the operating routine.

Execution across sites

Execution across sites worked only when the method was consistent enough to compare and flexible enough to respect exposure. A food operation, a logistics site and a maintenance-intensive plant can share the same 5 baseline dimensions, although the field questions under each dimension should reflect the work.

The first pass usually maps the declared system. That includes ISO 45001 clauses, local legal duties, internal standards, audit criteria, training matrices and committee routines. The second pass tests the lived system, which means walking the job, asking supervisors about recent trade-offs, checking closed actions and listening for the language workers use when they describe production pressure.

The third pass compares the two. When a site says stop-work authority is strong but workers describe 3 informal approvals before a job can stop, the baseline has found a cultural control gap. When a plant reports 98% training completion but supervisors cannot explain the control they are expected to verify, the baseline has found training theater.

ILO guidance on occupational safety and health management systems emphasizes worker participation because controls depend on people who can report defects before harm occurs. In a baseline, that principle becomes practical. The question is not whether participation exists on paper, but whether a worker can raise an inconvenient signal and receive a visible response.

What did the measured result prove?

The measured result proved that culture work becomes credible when it connects to operating indicators leaders already respect. During Andreza Araujo's PepsiCo South America tenure, the accident ratio fell 50% in 6 months under a 180-day plan, a result that shows why culture cannot stay trapped in slogans.

That PepsiCo result should not be copied as a promise. It should be read as a governance lesson. The reduction depended on leadership cadence, supervisor ownership, field verification and disciplined follow-up, which is why a related article analyzes the 3 safety culture decisions behind the 50% result.

50% in 6 months is not a culture slogan, it is a signal that the organization connected diagnosis with daily control. The baseline matters because it identifies where that connection is missing before leaders spend another year funding campaigns that do not change work.

As Andreza Araujo argues in Safety Culture: From Theory to Practice, culture is cultivated through repeated decisions. A baseline therefore has to show which decisions are repeated. If every site says safety is a value but only 2 of 8 sites can prove how managers protect safety under production pressure, the average score is hiding the real work.

Why did some sites resist the baseline?

Some sites resisted the baseline because evidence threatens the comfort of reputation. A plant with good audit history may not want worker interviews, field observations or action-aging reviews to expose how much of its safety performance depends on individual effort rather than a controlled system.

The first resistance pattern is defensiveness from high-scoring sites. They treat the baseline as a challenge to past performance, even though the purpose is to find drift before it becomes harm. The second pattern is overload from weaker sites, where leaders fear that every finding will become another action plan with no resources. The third pattern is translation loss, where corporate language does not match the words workers use locally.

These traps explain why the baseline must be led as a business decision, not as an EHS inspection. If the plant manager delegates every finding back to the safety department, the work teaches the organization that culture still belongs to EHS. That mistake is closely related to the ownership problems discussed in safety ownership myths plant managers must stop.

Generalizable lessons

The generalizable lessons from multi-site baseline work are practical because they apply even when the company has no global transformation program. The first lesson is that averages are politically convenient and operationally dangerous. They let leaders say the culture is improving while hiding the 1 site, 1 shift or 1 contractor group where risk is escaping.

The second lesson is that evidence must travel upward. If workers report weak signals but supervisors filter them, the executive team sees a sanitized culture. If supervisors escalate but managers reward only production recovery, the site learns that escalation is tolerated only when it does not disturb the plan.

The third lesson is that every baseline needs a decision clock. Findings that sit for 90 days become proof that the diagnosis was symbolic. Findings that are reviewed in the first 7 days, assigned to named owners and checked in the field teach the organization that the baseline has authority.

Each month without a multi-site culture baseline allows local drift to become normal, while leaders continue comparing sites through numbers that may not describe how risk is actually controlled.

What should an EHS manager apply in the next 30 days?

An EHS manager should apply a narrow baseline in the next 30 days by choosing 3 sites, 4 evidence streams and 1 executive decision forum. The goal is not to diagnose everything. The goal is to prove whether the organization can see cultural differences before injury data reveals them.

Start with the question leaders avoid. Which site looks strong on paper but worries experienced supervisors? Then test that concern through records, field observations, interviews and closed action quality. If all 4 evidence streams point to the same weakness, the baseline has found a decision issue, not a communication issue.

3 sites, 4 evidence streams and 30 days are enough for a pilot when the scope is disciplined. A pilot should not produce a thick report. It should produce a short decision memo that identifies where ownership is strong, where it is fragile and which leadership routine must change first.

For practitioners who want to deepen the method, Safety Culture Diagnosis gives the practical foundation for turning culture perception into evidence. The strongest baseline does not ask leaders to believe in culture. It shows them where culture is already deciding risk.