What 2,300 Behavior-Based Safety Projects Reveal About Leadership Quality

A safety leader can show a board thousands of completed observations and still be unable to answer the question that matters most: what changed in the work because people spoke up?

That question sits at the center of a five-year review of 2,300 behavior-based safety projects discussed by Dr. Thomas Krause on the Headline Podcast. The strongest predictor of success was not the number of workers reached. It was the quality of leadership given to the initiative. That finding changes the case for behavior-based safety from a frontline technique into a leadership test.

Initial scenario: the program looked active, but the result was uneven

Behavior-based safety programs often begin with visible energy. Workers receive observation cards or digital forms, supervisors review trends, and leaders announce that everyone owns safety. The activity can look substantial within weeks, especially when the organization rewards completion counts.

The problem appears when the program encounters a condition that a worker cannot fix alone. A machine guard is difficult to use during changeover. A contractor arrives without the right equipment. A production target compresses the time available for isolation. A supervisor receives the same concern across several shifts, yet the response remains a reminder to be more careful.

At that point, the program reveals its real design. If observations are treated as evidence about worker discipline, the initiative becomes a reporting layer that sits on top of unchanged work. If observations are treated as evidence about how work is led and designed, they can guide decisions about equipment, staffing, authorization, maintenance, and priorities.

Dr. Krause's finding matters because it places responsibility where the organization has the authority to act. Workers can identify a barrier and describe how the task behaves under pressure. Leaders decide whether the barrier receives time, money, engineering attention, or a change in the operating plan.

The decision: leadership quality had to become the central variable

The review did not make behavior-based safety valuable by adding another checklist. It made the leadership relationship visible. The same method can produce very different outcomes depending on whether leaders listen, remove obstacles, protect the process from punishment, and return to the field to verify what changed.

This is consistent with Andreza Araujo's experience across more than 250 cultural transformation projects. A safety initiative gains credibility when the organization responds to inconvenient information with a decision, not with a slogan. The worker does not need every concern solved immediately, but the worker does need to see that the concern entered a credible path toward assessment and action.

The leadership decision is therefore specific. Treat the program as an operating system for improving work, then assign executives responsibility for the conditions that frontline teams cannot control. Do not ask employees to own safety while withholding the authority required to change the task.

Execution: what high-quality leadership changes in the field

Leadership quality becomes observable through ordinary conversations. A supervisor who asks, “Why did you do that?” may receive a defensive answer. A supervisor who asks, “What made the safer option difficult here?” is more likely to learn whether the control was available, usable, and compatible with the real task.

That distinction does not excuse unsafe conduct. It improves the diagnosis. James Reason's work on latent failures helps explain why the final action is often shaped by earlier decisions about design, supervision, workload, maintenance, and communication. A useful observation process keeps both parts in view, the behavior that occurred and the conditions that made it more likely.

In a mature execution model, frontline employees help define what a good observation looks like, choose the conversations that matter, and identify recurring barriers. Leaders then do the work that only leaders can do. They set priorities, resolve conflicts between production and protection, fund corrections, and make ownership visible when an action crosses department boundaries.

The process also needs protection from the “gotcha” dynamic described in the Headline Podcast grounding. When an observation becomes a trap, workers learn to report what is safe to report. When the conversation helps the team meet the standard in operationally sound ways, people are more willing to reveal the gap between written expectations and working conditions.

Measured result: activity counts were not enough

The 2,300-project finding gives executives a useful warning about measurement. A high observation count can mean that the workforce is engaged, but it can also mean that the organization has optimized the easiest part of the process. Counting forms is simpler than proving that a recurring exposure disappeared.

A program should keep activity measures, because they show reach and cadence. They should not be mistaken for proof of effectiveness. The stronger evidence sits in the chain between signal and changed condition. A concern is reported, the decision owner is named, the control is changed, and the field verifies that the change survives the way work is actually performed.

This is where leadership quality becomes measurable without reducing it to a personality score. Review response time, action quality, closure evidence, repeat signals, worker participation in verification, and the number of issues that required escalation beyond the supervisor's authority.

Generalizable lesson one: give ownership to the people closest to the work

Frontline ownership is not the same as transferring responsibility downward. People closest to the work often know which control fails during a handover, which instruction conflicts with the equipment, and which shortcut appears whenever the schedule tightens. Their knowledge should shape the program.

Ownership grows when employees can influence the questions, the observation method, and the improvements that follow. It weakens when the program is imposed as a surveillance tool whose main purpose is to classify workers as compliant or noncompliant.

The executive implication is straightforward. Ask whether workers help design the process, then ask whether the process has a safe route for information that contradicts management assumptions. If the answer to either question is no, participation numbers will not repair the design.

Generalizable lesson two: leaders must remove obstacles they did not create

A worker can choose to use a control, yet still face a work system that makes the control slow, awkward, unavailable, or incompatible with the task. A supervisor can coach the behavior and still lack the authority to repair the condition. The leadership team must recognize that gap before it turns into repeated exposure.

During review meetings, ask which barriers require a decision outside the local team. The answer may involve procurement, engineering, maintenance planning, contractor governance, staffing, or production sequencing. Naming the barrier is not enough. Assign the decision to the role that controls the relevant resource and set a verification date.

This is also where safety leadership separates itself from motivational messaging. A leader does not demonstrate commitment by repeating that safety is the highest priority. A leader demonstrates it by accepting the operational consequence of a decision that protects people.

Generalizable lesson three: verify improvement in the task, not only in the dashboard

Closure is credible only when the field can demonstrate the changed condition. The team should be able to explain what is different, why the control is usable, what happens when the normal process is disrupted, and who responds if the control becomes unavailable.

Verification should include the people who raised the concern, because a document review can miss the friction that reappears during cleaning, changeover, night shift, maintenance, or contractor handoff. Their participation is not a courtesy. It is a test of whether the improvement belongs to the work or only to the management system.

Repeat observations are especially valuable here. They can show that the organization closed the record without closing the exposure. When the same signal returns, leaders should review the decision and the conditions around it rather than simply asking why employees keep making the same report.

What executives should apply in the next 30 days

Start with a sample of recent observations that leaders consider successful. Trace each one from the original conversation to the final action. Note the decision owner, the elapsed time, the barrier that changed, and the evidence used to close the item. Then select a small group of recurring signals and repeat the review across shifts or sites.

Use the review to ask four questions. Did the worker have a realistic way to follow the expected method? Did the supervisor have authority to protect the task? Did the organization address the condition that made the deviation likely? Can the team demonstrate that the change works when production pressure returns?

Andreza Araujo's work brings the conversation back to culture. Across 25+ years of multinational EHS leadership, she has repeatedly emphasized the gap between declared commitment and operating reality. A behavior-based safety program is useful when it exposes that gap early enough for leaders to close it.

For more conversations on the decisions that shape safer workplaces, visit the Headline Podcast, the space where leadership and safety come together to shape better workplaces and better lives.

The case in one sentence

The review of 2,300 behavior-based safety projects points to a durable conclusion: frontline observations create value only when leadership quality converts them into better work, accountable decisions, and verified control changes.