Normalization of Deviance: 7 Signals Supervisors Miss

NASA's Columbia Accident Investigation Board described normalization of deviance as a pattern in which repeated exceptions slowly become accepted as normal work. This article gives supervisors and EHS managers seven field signals that show when a team has stopped treating drift as information and started treating it as proof that the risk is acceptable.

Why normalization of deviance is a supervisor problem

Normalization of deviance appears when a weak practice survives long enough to look legitimate. The operator skips one verification, the crew works around one missing guard, the supervisor accepts one incomplete briefing, and after several quiet weeks the shortcut becomes part of the local method.

The difficult point is that the first deviation rarely looks dramatic. Across 25+ years leading EHS in multinational environments, Andreza Araujo identifies that safety drift usually begins as adaptation under pressure, not as open defiance. That is why supervisors need to see repeated exceptions as cultural data, not only as disciplinary events.

In safe-behavior work, the strongest question is not whether people know the rule. The stronger question is whether the operation rewards the rule when schedule, peer pressure, fatigue, and production recovery all move in the opposite direction.

1. The exception has become easier to explain than to stop

A deviation is becoming normalized when the explanation arrives faster than the correction. In many operations, the sentence sounds harmless: the guard is off because maintenance is still adjusting the line, or the pedestrian route changed because the materials were staged late.

What most safety conversations miss is that fast explanation often protects the local system from scrutiny. As Andreza Araujo argues in Safety Culture: From Theory to Practice, culture is revealed less by what leaders declare and more by what they tolerate under operational pressure.

The supervisor should ask three questions before accepting the explanation: who benefited from this workaround, how many times it happened in the last thirty days, and what condition made the correct behavior harder than the unsafe one. If the answers repeat, the issue is no longer a one-off behavior problem.

This is where risk perception drift becomes visible, because the team has started to judge danger by recent luck rather than by the energy, exposure, and consequence still present in the task.

2. Pre-task briefings describe the task but not the drift

A pre-task briefing is weak when it names the official hazards but ignores the actual deviations present at the worksite. The form may mention working at height, mobile equipment, or hazardous energy, while the crew already knows that the anchor point, route, tool, or sequence has changed.

The Columbia investigation showed how repeated acceptance can turn warning signs into background noise. In a plant, warehouse, mine, or construction site, the same pattern appears when the crew treats today's exception as familiar instead of abnormal.

A supervisor should make the briefing expose the gap between the plan and the field. One practical method is to ask, "What is different from the last time we did this work?" and then require the crew to name the control that changes because of that difference.

If the briefing cannot identify drift, it becomes ceremony. The stronger alternative is a pre-task briefing that catches behavior traps before the team enters the highest-exposure part of the job.

3. Near misses are counted but not classified by drift

Near-miss reporting loses preventive value when all events enter the same bucket. A dropped tool, a bypassed interlock, a missed isolation point, and a shortcut around traffic segregation do not teach the same lesson, even if none produced injury.

In more than 250 cultural transformation projects, Andreza Araujo observes that weak reporting systems often measure volume while missing quality. 250+ transformation projects give a clear lesson: the best signal is not the number of reports, but whether reports reveal repeating conditions before those conditions become accepted practice.

The corrective move is to add a simple drift marker to near-miss classification. The reviewer should flag whether the event involved a repeated workaround, an accepted exception, a missing barrier, peer pressure, or schedule pressure.

That single classification change turns near-miss quality into a leading indicator. Without it, the dashboard may look active while the operation quietly teaches people which rules are negotiable.

4. Stop-work authority exists on paper but feels socially expensive

Stop-work authority is not real when the person who uses it pays a social price. The procedure may say that every worker can stop unsafe work, yet the crew may punish the interruption through sarcasm, exclusion, pressure, or the label of being difficult.

This matters because normalization of deviance often survives through group approval. If the team has completed the same unsafe workaround ten times without consequence, the worker who challenges it becomes the disturbance, although the true disturbance is the drift itself.

The supervisor should test the system after every stop-work event. Ask whether the person was thanked in public, whether the work plan changed, and whether the interruption exposed a condition that leadership had ignored. If nothing changes after the stop, the authority is symbolic.

A mature stop-work authority design protects the person who interrupts drift and forces management to remove the condition that made interruption necessary.

5. The team uses good results as evidence that the shortcut is safe

A shortcut becomes dangerous when recent success is used as proof of safety. The team may say that the method has worked for months, even though the energy source, fall exposure, chemical release potential, or vehicle interface has not changed.

James Reason's work on latent failures helps explain the trap. A system can absorb weak decisions for a long time, and that quiet period can deceive people into believing the barriers are stronger than they are.

The supervisor should separate outcome from exposure in the daily conversation. A task that ended without injury may still have operated with a missing barrier, and that missing barrier should be recorded as a serious condition rather than dismissed as an uneventful shift.

50% accident reduction in six months during Andreza Araujo's PepsiCo South America tenure shows why this distinction matters. The result came from changing how risk was governed, not from celebrating quiet weeks while weak practices stayed untouched.

6. Fatigue and production recovery make the drift predictable

Normalization of deviance accelerates when fatigue, overtime, or production recovery makes exceptions predictable. The risky practice may appear every Friday night, after a line stoppage, during shutdown recovery, or when a supervisor is covering two areas at once.

The Headline Podcast editorial lens treats this as leadership evidence rather than worker weakness. When the same pressure pattern produces the same unsafe adaptation, the leader has enough information to redesign the work, staffing, timing, or decision rights.

The practical test is to map deviations by time, crew, task phase, and production condition. If most deviations cluster after extended hours or recovery work, the countermeasure should not be another reminder poster. It should be a change in planning, supervision density, or pause criteria.

Decision load also matters. A supervisor facing constant tradeoffs between speed and control may need a clearer trigger for escalation, especially in environments where decision fatigue turns weak choices into routine choices.

7. Corrective actions retrain people but leave the drift source intact

Retraining is a weak corrective action when the deviation was created by a work-system condition. If the task sequence, tool availability, staffing model, layout, or approval flow still pushes people toward the workaround, the next person will repeat the same behavior with fresh training records.

Andreza Araujo's book A Ilusao da Conformidade, often translated as The Illusion of Compliance, is useful here because it separates documented conformity from operated conformity. A signed form can coexist with a field practice that everyone knows is different.

The supervisor should require every corrective action to name the drift source. Was it time pressure, missing equipment, unclear authority, poor access, conflicting goals, or peer acceptance? Only after that answer is clear should training be considered part of the solution.

Each month without this classification allows the operation to normalize more exceptions, while leaders receive cleaner records than the field reality deserves.

Comparison: isolated violation vs normalized deviance

Conclusion

Normalization of deviance is not a label for careless workers, but a warning that the operation has started to convert repeated exceptions into accepted practice.

Supervisors can break the pattern by treating drift as data, asking where the correct behavior became difficult, and escalating conditions that training alone cannot repair. For more leadership conversations on safety, culture, and risk, follow Headline Podcast.