Speak-Up Systems: 6 Failure Points That Make Bad News Arrive Too Late

A worker raises a concern about a lifting plan that no longer matches the equipment on the floor. The supervisor thanks the worker, opens a ticket, and promises to review it. Two shifts later, the same lift is still scheduled, the worker has heard nothing, and the concern has become a private warning instead of an operational decision.

This is where many speak-up systems fail. They make it easier to submit information without making it harder for the organization to ignore information. The central question is not whether workers can speak. It is whether a credible concern reaches a decision owner soon enough to change the exposure.

Psychological safety, as Amy Edmondson describes it, gives people permission to take interpersonal risks such as asking a question, admitting uncertainty, or challenging a plan. A safety speak-up system must add the operational layer that turns that permission into protection. It needs a clear route from voice to evidence, from evidence to decision, and from decision to field verification.

Why can a speak-up system look active while serious concerns still arrive too late?

Reporting volume is an incomplete measure. A site may receive hundreds of observations because its software is easy to use, while workers withhold the concerns that could challenge a supervisor, interrupt production, or expose a design weakness. The visible stream then contains low-consequence items, while the most important information travels through private conversations, informal messages, or not at all.

The delay often sits between the first report and the first accountable decision. Someone receives the concern, but that person lacks authority. A manager owns the area, but does not know the escalation threshold. The EHS team can advise, although it cannot control the schedule or fund the correction. Each handoff appears reasonable, yet the exposure continues while responsibility moves.

James Reason's work on active and latent failures is useful here because the failure is rarely located in the worker's willingness to speak alone. The organization can create latent conditions through unclear ownership, weak interfaces, and incentives that reward rapid closure more than accurate escalation.

Failure point 1: The channel collects a message without defining a decision

A form that asks for the hazard, photograph, and location may create a clean record, but it does not answer the operational question. What decision must follow this concern? The task may need to stop, a temporary control may need to be installed, a technical review may need to begin, or the concern may require monitoring rather than immediate interruption.

When the system does not define the possible decisions, recipients default to administrative action. They assign an owner, set a due date, and close the record when a comment appears. The process becomes complete even though the risk question remains open.

Repair this failure by requiring each serious concern to name the decision in plain language. A useful record should state what could happen, which condition is uncertain, who can authorize the next step, and what evidence would justify continuation. That design keeps the report connected to work rather than to a queue.

Failure point 2: The escalation threshold is known only by experienced insiders

Workers often learn escalation rules through observation. They notice which supervisor accepts a challenge, which department delays a response, and which types of concern receive attention. New employees, contractors, and temporary workers do not have the same informal map, so they may interpret silence as a signal that the issue is not serious enough to raise.

A reliable system makes the threshold visible before a person needs it. The threshold should consider potential consequence, exposure frequency, uncertainty about the control, and the organization's ability to recover if the task proceeds. It should not depend on the speaker having the vocabulary of an EHS specialist.

Supervisors can make the threshold practical by asking, “What would make this concern urgent?” during pre-job conversations. The answer should identify the condition that changes the decision, not simply repeat a general rule about reporting hazards.

Failure point 3: The first recipient is expected to solve a problem outside their authority

Many organizations tell employees to speak to their supervisor first. That is sensible when the supervisor can change the work, but it becomes unsafe when the supervisor is part of the pressure that created the exposure or cannot authorize the needed correction.

The person who receives the concern needs a defined responsibility that is smaller and clearer than “solve it.” Their first duty is to protect the signal. They should acknowledge the concern, assess whether immediate protection is needed, identify the decision owner, and preserve the connection with the person who raised it. That sequence prevents a frontline recipient from quietly absorbing a problem they cannot resolve.

Leaders should test this path with real scenarios. Ask a supervisor who can stop a contractor lift, who can release engineering support, who can change a production sequence, and who must be notified if those options are unavailable. If the answers conflict, the speak-up system has a decision-rights gap.

Failure point 4: Feedback is treated as courtesy instead of control verification

Thanking a worker for speaking is important, although gratitude is not feedback. Feedback explains what the organization understood, what decision it made, what protection applies, and when the person can expect another update. Without those elements, the worker cannot tell whether the concern changed the work.

The absence of feedback creates two distortions. People who raised concerns may stop doing so because they believe reporting is performative. Managers may believe the system is healthy because the database contains completed actions. Both groups lose sight of the original condition.

A simple case-close test can correct this. Before closing a serious concern, the responsible leader should confirm that the speaker received an explanation, that the control was checked where the work happens, and that any remaining uncertainty has an owner. This is the practical meaning of participation in a safety system.

Failure point 5: Leaders reward the number of reports and the speed of closure

Metrics shape behavior before a policy does. If supervisors are praised for reducing open items, they have a reason to classify complex concerns as routine tasks. If sites compete on reporting volume, workers have a reason to submit safe observations that are easy to close. Neither metric is useless, but either can become misleading when treated as proof of psychological safety.

Better measures follow the quality of the decision path. Track how long a high-consequence concern takes to reach an accountable owner, how often the initial control remains unresolved, how many concerns recur after closure, and whether field verification confirms the promised change. Review a sample of cases with the person who raised the concern, not only with the person who closed it.

Across more than 250 cultural transformation projects, Andreza Araujo has emphasized the difference between visible activity and changed conditions. A dashboard that celebrates closure while the same exposure returns is measuring administrative motion, not control reliability.

Failure point 6: The system protects the report but not the relationship

Workers do not evaluate a speak-up system only through its software. They evaluate it through the next conversation. A supervisor who asks, “Why did you wait until now?” may not issue a formal punishment, yet the question can make future silence rational. A manager who treats a concern as disloyalty can damage trust across a whole shift.

Psychological safety does not mean every challenge is correct or that leaders remove standards. It means people can present information, uncertainty, and disagreement without being humiliated for doing so. Leaders still examine evidence, correct unsafe choices, and hold people accountable for deliberate violations. They must separate the quality of a concern from the social status of the person who raised it.

Andreza Araujo develops this distinction in *A Ilusão da Conformidade*, which challenges the assumption that a documented process is working simply because people appear to follow it. The same test applies to speak-up. If people must calculate the personal cost of raising inconvenient evidence, the system is compliant on paper and weak in practice.

What should leaders test in the next 30 days?

Start with five recent concerns that involved a changed condition, a high-consequence exposure, or a disagreement about whether work could proceed. Reconstruct the path from the first words spoken to the final decision. Record where the concern waited, which handoff created ambiguity, and whether the person who raised it received a meaningful explanation.

• Ask whether the concern named a decision, not only a hazard.
• Confirm that the first recipient knew the escalation threshold and the decision owner.
• Check whether temporary protection was defined while evidence was incomplete.
• Interview the worker who raised the concern and compare their account with the database record.
• Verify in the field that the promised control changed the exposure.

Then repair one interface instead of launching another campaign. Clarify who can stop the work, who must be notified, what response time applies, and how the person who spoke will be updated. A narrow repair that changes a real decision is more valuable than a broad message about the importance of speaking up.

How does a mature speak-up system change safety leadership?

It changes the leader's job from encouraging voice in the abstract to designing a dependable response under pressure. Leaders must make it safe to raise the concern, easy to identify the decision owner, and difficult to close the record without checking the condition. That work requires attention to interfaces because serious exposure often survives between departments rather than inside one person's task.

During her tenure at PepsiCo South America Foods, where she led a 50% reduction in the accident ratio in six months under a 180-day plan, Andreza Araujo treated safety performance as a management system rather than as a communication campaign. The lesson for speak-up is direct. Voice becomes a control only when leaders connect it to a decision, a resource, and evidence that the condition changed.

As *Safety Culture: From Theory to Practice* argues, culture becomes visible through repeated choices. A worker who speaks up should be able to see what choice followed. That visible link is how psychological safety becomes operational safety.

What is the clearest test of speak-up credibility?

The clearest test is a difficult concern raised before an incident, followed by a response that protects the person, clarifies the uncertainty, assigns decision authority, and verifies the control in the field. If the organization can perform that sequence consistently, reporting is connected to prevention.

If it cannot, the answer is not another slogan or a larger reporting target. Find the point where information loses ownership, then repair that handoff. The safety of speaking up depends on what the organization does next.