Publication date: October 2026
Fifteen minutes beside the production line, a signed attendance sheet, and an objection that almost no one records. This article looks at why safety culture programs fail when they are introduced in the wrong order—and what the evidence says about the right sequence.
The micro-training lasts 15 minutes and takes place beside the production line, with the press running in the background and half the shift on the floor. Someone from safety explains the correct procedure, shows two photos from the corporate newsletter, asks if anyone has questions, and passes around the attendance sheet. The signatures take two minutes to collect. The record is flawless.
And then, almost always near the end, after the form has been signed and the conversation has turned informal, someone says it:
—That’s fine. I’ll do it. But with what?
I don’t have the tool. My supervisor tells me to do it this way or I won’t meet the production rate. The guard has been stuck for months. We reported it at the time.
That sentence does not appear in any metric. There is no box for it on the attendance sheet. It does not make it into the monthly training report or the safety committee meeting. And yet, technically, it is the most precise objection raised during the entire session.
It deserves to be taken seriously. It is not resistance to change; it is a diagnosis, coming from the only place in the system where someone can see the instruction, the available resources, and the cost of missing the target—all at the same time.
The operator is pointing out three things: they were asked to behave a certain way without being given the resources to do so; they are receiving conflicting instructions from two different sources; and the channel they used to report the problem produced no visible result.
Then there is the supervisor. It is easy to cast the supervisor as the story’s antagonist, but that is not the case. The person asking for a certain pace is passing along an instruction they received within a measurement system that rewards one variable and barely mentions the other.
Zohar showed that the frontline supervisor acts as a translator: turning the policy stated by leadership into the day-to-day climate the team experiences. The UK Health and Safety Executive (HSE), for its part, identifies supervision as a contributing factor in major accidents when supervisors lack the resources, clear responsibilities, or time to do their job.
No one in this scene is acting in bad faith. The system design is failing.
Over the past decade, safety culture has become the stated focus of industrial safety in manufacturing. Along the way, it has become intertwined with other legitimate priorities: inclusion, corporate social responsibility, employer branding, and workplace certifications. None of those is the problem.
The UK HSE makes a point that is worth citing because it runs counter to the prevailing assumption. Companies, it says, often talk about safety culture in the narrow sense of employee compliance, when management culture is what matters. Its inspectors’ handbook lists three indicators of a poor safety culture. None focuses on the operator:
The same agency identifies six major influences on safety culture: management commitment and style, employee involvement, training and competence, communication, compliance with procedures, and organizational learning. Four of the six are squarely management’s responsibility.
The practical implication is rarely drawn. If culture is largely shaped by management decisions, a program that holds operators and team leaders accountable before managers is measuring the effect, not the cause.
The Bradley Curve, developed by DuPont, is the industry’s best-known model, and for good reason: it accurately describes four states anyone who has worked in a plant will recognize—reactive, dependent, independent, and interdependent. As a conversation tool for leadership, it works.
Its limitation is different: it describes states without explaining how to move from one to the next. In practice, that gap is often filled with a dangerous assumption. If the goal is interdependence—looking out for one another—then the starting point must be asking operators to take responsibility for themselves and their coworkers.
The final stage thus becomes the starting technique.
There is a model with a similar structure and a published research foundation. Developed by the Keil Centre for the HSE, the Safety Culture Maturity Model defines five levels—emerging, managing, involving, cooperating, and continually improving—and establishes that organizations progress sequentially, building on the strengths and addressing the weaknesses of each preceding level. Its key contribution is an operational warning:
Safety improvement techniques can fail when they do not match the organization’s level of maturity.
So safety culture programs do not fail because of a lack of commitment, employee indifference, or poor communication. They fail because the tool does not match the organization’s stage. That is good news: a mismatch can be corrected through assessment, not enthusiasm.
The levels shown are based on the Keil Centre model for the HSE. The assignment of techniques to each stage is a professional interpretation of the model’s matching principle.
The industry uses two terms as if they were interchangeable, but research carefully distinguishes between them. Safety culture is the broader concept: shared assumptions, values, and observable practices. Safety climate is its measurable expression at a given moment—the workforce’s perception of the organization’s actual priorities.
The distinction matters because safety climate is measured, while safety culture is invoked. Since Zohar’s foundational work, more than 200 indexed articles on safety climate have been published, and more than 20 measurement tools have been validated. A meta-analysis of 32 studies found a moderate correlation between safety climate and both safe behavior compliance and safety participation. It also found an inverse correlation between safety climate and injuries, stronger in prospective than in retrospective studies.
There is another finding that sheds light on the entire discussion: organizations with a poor safety climate significantly underreport injuries. This means the actual relationship between safety climate and injury rates is probably underestimated in the research.
Read that sentence slowly while looking at the “days without an accident” board that hangs at the entrance of nearly every plant. Technically speaking, that number creates a social and financial incentive not to report. No one wants to be the person who resets it to zero. A plant with an improving safety climate will often see its reports increase for a while. A manager who celebrates a low number may be pushing the culture in the opposite direction from the one they intend.
There is an unproductive debate about behavior-based safety between its supporters and its critics. The HSE’s position is more useful than either side’s because it is based on sequence, not ideology.
The agency says a behavior-based program should not be introduced unless the organization is already confident that it has a sound management system and a safe workplace.
It adds that such a program can be a useful addition to occupational safety, with limited benefits for controlling major hazards. It can divert attention from management system fundamentals and process safety. A failed program can leave an organization worse off than before, and it must not shift responsibility from management to the individual.
This is not a ban. It is a matter of order. Behavioral observation belongs at the more mature levels, when the system it is meant to observe is already in place. Introduced earlier, it has no system to support it.
Here is the least discussed consequence, and probably the most serious.
A micro-training session that informs an operator about a hazard, assigns them responsibility for controlling it, and changes neither the tool, the method, nor the production target creates a specific result: someone who now knows what they should do but still does not know what they can use to protect themselves. Awareness has increased; the ability to act has not.
Occupational health research has a name for this combination. High psychological demands combined with low decision latitude is the operational definition of job strain. Its effects on blood pressure, cardiovascular risk, musculoskeletal disorders, and mental health are documented. Responsibility without authority is not a figure of speech; it is a measured and recognized psychosocial risk factor.
A poorly sequenced safety culture program, then, is not simply ineffective. It can itself become a psychosocial hazard. In Mexico, this is not merely an academic consideration: NOM-035-STPS-2018 requires employers to identify psychosocial risk factors and promote a favorable organizational environment. A workplace where reporting carries a personal cost, and where employees are held responsible without being given the means to act, falls within the scope of that standard.
The solution is well known, and it is not new. Occupational health psychology uses the same intervention framework as industrial hygiene: the hierarchy of controls. Primary prevention eliminates or reduces the stressor at its source—the job design, pace, workload, or missing tool. Secondary prevention changes how people respond to the stressor. Tertiary prevention addresses harm that has already occurred. NIOSH has emphasized focusing efforts on primary prevention of organizational risk factors. Almost everything industry buys as a “culture program” is secondary prevention disguised as primary prevention.
On paper, the theory sounds flawless. But when this illusion of safety collides with real pressure on the production floor, the metrics begin to lie.
To understand this illusion, imagine a stamping and assembly plant in the automotive corridor that has just reached 900 days without a lost-time injury. The entrance board announces it in large digits. The behavior observation program has been running for two years, with a quota of 12 cards per supervisor each month, and the quota is met. Micro-training sessions are held every two weeks, and training compliance is 98%.
The latest quarterly report shows three near misses.
Read alongside the other numbers, that third figure does not describe a safe plant. It describes a silent one. And the silence broke during a die change, a nonroutine task, when an interlock that had been tampered with months earlier stopped working. The subsequent investigation uncovered three informal mentions of the problem. None became a report because the reporting channel existed on paper, not in practice.
The HSE calls this passive learning—identifying lessons without implementing them—and describes the antidote with two expressions worth adopting in Spanish: chronic unease, actively looking for information instead of waiting for a report; and amplifying weak signals, listening carefully to minor concerns raised by frontline workers.
What should be done—and what went wrong
None of this means the plant did too little. It did a lot, with discipline and resources. It did a lot in the wrong order.
The question that opens this article—“But with what?”—has an answer. This is the constructive part: conditions and consequences change under identifiable circumstances, and none of them is a campaign.
That last point defines what industrial safety leadership means today, and it looks nothing like the list of virtues it is often reduced to. Zohar showed that leadership affects injuries through two channels: the organizational climate and supervisors’ practices, which translate policy into action on the floor. The HSE translates this into observable behaviors: visible leadership, regular presence on the production floor, and a willingness to stop production to address a safety issue.
Stopping the line communicates more than any campaign. It is the only unambiguous way to show which of the two priorities governs when they conflict.
It is worth returning to the micro-training session beside the line, because it can end differently.
It ends differently when an operator asks what tool they should use and there is a clear answer: the tool is available, the method was reviewed with the operator before it was written, and the production target was adjusted to accommodate the change. It ends differently when the person raising a concern knows it will go through a channel that produces visible results—and that speaking up will not cost them anything. It ends differently when the supervisor does not have to choose between two conflicting instructions because leadership has already made the choice.
Awareness stops being a burden and becomes what it should have been all along: the final reinforcement of a system that already provided protection. That is the right sequence. In industrial safety, sequence matters more than almost anything else.
If your safety culture program is not moving the metrics and you suspect your organization’s maturity does not match the tool you are using, let’s talk.
At HF Safety & Environmental Services, we help manufacturing, food processing, pharmaceutical, and chemical plants get back on track—from assessing the safety climate and verifying prerequisites to developing frontline leadership that works in practice, not just on paper.
Does behavior-based safety work?
It can, in the right place and at the right time. The HSE considers it a useful addition to occupational safety when the organization already has a sound management system and a safe workplace. It also cautions that its benefits for controlling major hazards are limited. Introduced before those prerequisites are in place, it can leave the organization worse off than before.
Is the Bradley Curve useful?
Yes, as a descriptive map: it accurately names recognizable states. Its limitation is that it does not explain how to move from one state to another. The Keil Centre’s maturity model for the HSE addresses that gap by establishing that improvement techniques can fail if they do not match the organization’s level of maturity.
Why do reports of unsafe conditions and behaviors increase when the safety culture improves?
Because reporting no longer carries a personal cost. Research on safety climate documents significant injury underreporting in organizations with a poor safety climate. An increase in near-miss reports is often a sign of improvement, not deterioration.
How does this relate to NOM-035-STPS-2018?
Holding people responsible without giving them the means to act creates high demands and low decision-making authority, a condition associated with documented health effects. A workplace where reporting carries a personal cost falls under the standard’s requirements to identify psychosocial risk factors and promote a favorable organizational environment.
Todos los Derechos Reservados 2026