Unilever EHS Governance: How 19 Countries Aligned Risk

During Andreza Araujo's tenure as Director SHE LATAM at Unilever, the EHS scope covered 19 countries, 30,000 employees, 34 factories and more than 60 distribution centers. This case study shows why multi-country risk governance works only when leaders standardize the language of risk without pretending that every site faces the same exposure.

The lesson is uncomfortable for global EHS managers because local autonomy can protect reality, while unmanaged autonomy can hide fatal variation. A register that means one thing in Mexico, another in Colombia and something else in Argentina is not a risk system, even if every site uploads a file on time.

Why 19-country EHS governance is a risk-management problem

Multi-country EHS governance is not mainly a reporting challenge. It is a translation challenge in which hazards, controls, accountability and escalation thresholds must mean the same thing across countries whose laws, unions, suppliers and operational habits differ.

ISO 31000 frames risk management around principles, a management model and process, but the standard does not make a forklift route in a crowded distribution center comparable to a chemical transfer in a factory. The organization must build that comparability through definitions, evidence and decision rights.

Across 25+ years leading EHS in multinationals, Andreza Araujo has seen that the weak point is rarely the existence of a risk register. The weak point is whether the register tells a senior leader where control is strong, where control is assumed and where control is missing.

The counter-thesis is simple enough to test: a multinational can have aligned reporting and still have fragmented risk governance. Reporting answers whether the form arrived; governance answers whether the right person saw the real exposure and had the authority to change the decision.

Initial scenario

The initial scenario in a regional EHS role is usually a mosaic. One country classifies contractor road risk as high because it has recent events, another classifies the same exposure as medium because the matrix score is averaged, and a third treats it as an operational issue outside the central EHS dashboard.

That variation looks like local maturity, but it can also become an invisibility mechanism. When the same exposure receives different names, different scores and different escalation rules, the regional team cannot compare risk quality across 34 factories and 60+ distribution centers.

In Andreza Araujo's Unilever LATAM scope, the managerial problem was not to erase local reality. It was to create a shared grammar for risk so that leaders could discuss exposure, control and residual risk without losing the operational detail that makes each site different.

The most dangerous version of this scenario appears when every site appears compliant. The archive is complete, the meeting cadence is visible and the action list is active, although the underlying controls do not prove that serious exposure has been reduced.

Decision

The decisive move in a regional governance model is to separate what must be common from what must stay local. Severity definitions, escalation triggers, critical-control criteria and evidence standards need a common spine, while the site-specific scenarios remain local because the hazard profile is not identical.

This is where many companies fail. They standardize the template and call it governance, although the template only proves that the same spreadsheet traveled across countries. Governance begins when the same exposure produces the same leadership question, even if the corrective action differs by site.

As Andreza Araujo argues in *Sorte ou Capacidade* (Luck or Capability), well-managed risk is calculated and mitigated with method, not bravado. In a 19-country environment, method means that a serious exposure cannot depend on the confidence style of the local manager who presents it.

The decision also needs a refusal line. If a local team cannot show how a critical control is defined, verified and owned, the regional system should not accept the risk as controlled merely because the action owner says the work is in progress.

Execution

Execution starts with a taxonomy that reduces noise. Every site needs to classify operational risk using agreed categories such as energy isolation, vehicle movement, working at height, contractor interface, chemical exposure, process change and emergency response.

The next layer is evidence. A regional team should not accept a green status unless the site can show current control verification, field observations, closed actions and leadership review. This is why indicator triangulation matters: no single metric can prove that a risk is under control.

The third layer is escalation. If a site has repeated overdue critical actions, weak verification or a serious exposure accepted beyond its authority level, the issue moves from local tracking to regional review. That movement is governance, because it changes who has to decide.

Andreza Araujo's work across 250+ companies reinforces the same operating point: a risk system becomes credible when field evidence can challenge managerial confidence. Without that friction, the dashboard becomes a mirror for optimism.

Measured result

The measurable result in this case is not presented as an invented accident reduction number, because the verified public anchor is the scale of Andreza Araujo's Unilever LATAM responsibility: 19 countries, 30,000 employees, 34 factories and more than 60 distribution centers. The result that can be defended is the creation of a common risk conversation across that scale.

That matters because scale changes the nature of risk. A weakness that looks isolated at one factory may reveal a regional pattern when the same control gap appears in five countries, and a control that looks mature in one distribution center may fail when contractor turnover changes.

The strongest signal of maturity is not a prettier dashboard. It is the moment when a country manager, a plant leader and a regional EHS director can look at the same exposure and agree on whether the risk is accepted, transferred, reduced or escalated.

For a global EHS manager, this is a defensible result because it changes the quality of executive attention. The conversation moves from whether the site complied with a reporting cycle to whether the organization understands the risk it is carrying.

Before and after

This before-and-after view also explains why MOC, PTW and PSSR controls need different decision rules. A governance model that treats every control as a generic action item will miss the difference between approving a work permit and accepting a startup risk.

The table should not be read as a documentation upgrade. It is a decision-quality upgrade, because leaders can only prioritize scarce time, capital and attention when risk information has enough structure to support comparison.

Generalizable lessons

The first lesson is that risk governance must define authority before pressure arrives. If a plant manager can accept a high residual risk without regional review, the system has already decided that local production pressure outranks the common risk standard.

The second lesson is that visible risk is more useful than tidy risk. A messy dashboard that exposes weak control evidence is more valuable than a clean dashboard that turns every country green. This connects directly with the dock safety case, where visibility was the point of the governance work.

The third lesson is that culture and risk management cannot be separated. If people believe that escalation will be punished, they will sanitize the register; if leaders reward early exposure of weak controls, the register becomes a management tool rather than a compliance archive.

A fourth lesson sits beneath the first three: governance must protect the messenger. When local teams see that bad news produces support and clearer priorities, rather than blame, they report weak controls earlier and with better evidence.

What to apply in your operation

A global or regional EHS manager should begin with the five exposures that can kill, disable or create severe regulatory consequences. For each exposure, define the minimum control standard, the evidence required, the decision owner and the threshold that forces escalation.

Then test the system with three sites that are different on purpose: a mature factory, a pressured distribution center and a site with high contractor dependency. If the same risk language survives those three contexts, the model has a chance of working across the broader network.

The trap is to confuse participation with alignment. A country may attend every meeting and still use a different risk language, which is why safety culture drift often appears first in the way people classify, explain and minimize risk.

Andreza Araujo's practical lesson for this audience is that risk perception also lives in people, not only in tools. The right governance model therefore trains leaders to ask better questions, verify field evidence and refuse comfortable answers when the exposure is severe.

Conclusion

The Unilever LATAM case shows that regional risk governance is not a binder, a dashboard or a calendar of meetings. It is the disciplined act of making risk comparable enough for leaders to decide, while keeping reality specific enough for sites to act.

If your organization needs to turn fragmented site-level risk information into a governance model that leaders can actually use, Andreza Araujo's ACS Global Ventures supports diagnostics, roadmaps and implementation through Andreza Araujo.