Speak-Up Metrics: 7 Signals Leaders Should Track

Gallup's State of the Global Workplace 2024 report found that only 23% of employees were engaged at work, which means most organizations are trying to manage risk through a workforce that may not volunteer weak signals. This article shows how senior leaders can treat speak-up as a measurable safety indicator, not as a slogan that appears after an incident.

Why speak-up is a safety metric, not a personality trait

Speak-up becomes a safety metric when leaders measure whether people raise operational doubts, including workplace harassment concerns, before harm occurs.

Amy Edmondson's 1999 Administrative Science Quarterly article on psychological safety defined the team condition that allows people to take interpersonal risks. In occupational safety, that condition matters because the first warning of a serious event often appears as a question, a pause, or a refusal to accept a shortcut.

On the Headline Podcast, Andreza Araujo and Dr. Megan Tranter frame safety conversations as real conversations with constantly learning people. That stance changes the executive question. The issue is not whether employees are brave enough to speak, since courage is uneven by nature. The issue is whether the system makes speaking useful enough that rational people keep doing it.

For an EHS manager, the practical move, especially when middle manager burnout is creating reporting fatigue, is to place speak-up indicators beside SIF leading indicators, because both look for weak signals before the damage is visible in lagging metrics.

1. Track the time between concern and response

Response time measures how long it takes for a raised concern to receive a credible answer, owner, or decision.

A concern that waits three weeks for a reply teaches the workforce that silence is faster than reporting. ISO 45003:2021 asks organizations to manage psychosocial risks through the occupational health and safety system, and delayed response is one way the system creates frustration, mistrust, and emotional load.

Co-host Andreza Araujo has explored in Safety Culture: From Theory to Practice that culture is revealed in repeated managerial choices, not in declared values. If leaders say that voice matters but leave concerns in a queue with no owner, the spoken culture and operated culture separate.

Set a response service level by risk class. A high-potential fatal exposure should receive same-day triage, while a lower-risk improvement idea may have a seven-day response window. The metric is not speed alone, because a fast dismissal still kills trust. Measure whether the response includes classification, next step, accountable role, and expected date.

2. Separate volume from quality of voice

High reporting volume is not proof of psychological safety unless the reports include detail, dissent, and operational uncertainty.

Many dashboards count near misses, concerns, or observations as if more entries always meant a healthier culture. The trap is that a plant can flood the system with low-value observations while the dangerous doubts remain private. Gallup's 2024 report also reported 62% of employees as not engaged, and disengaged people often give minimum information even when they comply with reporting requirements.

The better indicator is voice quality. Count reports that challenge a plan, identify a design weakness, question a production deadline assumption, or name a missing barrier. This connects naturally with risk matrix blind spots, because the most useful voice often reveals what the formal matrix normalized too early.

Review a sample of reports each month and score them against three levels: descriptive, analytical, and decision-changing. The executive should care most about the third level, because those reports alter work design, resource allocation, or leadership behavior.

3. Measure who speaks, not only how often people speak

A speak-up system is weak when the same few confident people generate most of the safety voice.

The distribution of voice tells leaders whether psychological safety reaches contractors, new employees, night-shift teams, maintenance crews, and minority-language groups. If 80% of concerns come from supervisors and EHS staff, the dashboard is not measuring the workforce. It is measuring the safety function talking to itself.

In more than 250 cultural transformation projects, Andreza Araujo observes that participation patterns are often more revealing than survey averages. A site can score well on a climate question while operators with the highest exposure still avoid dissent in front of production leadership.

Build a participation map by role, shift, tenure, contractor status, and work area. Do not publish personal rankings, since that turns voice into surveillance. Use the map to find quiet zones, then send leaders to listen where the metric is low.

4. Audit what happens after dissent

Dissent only becomes a safety indicator when the organization tracks whether disagreement changes a decision.

Employees learn from consequences. If a mechanic questions an isolation plan and the supervisor treats the question as resistance, the next mechanic will probably stay quiet. If the team pauses, checks the energy source, and records the decision, dissent becomes part of work control.

This is where visible felt leadership must be more than a walkaround. The leader has to demonstrate that dissent can interrupt tempo without humiliating the person who raised it. The metric should capture decision changes, design reviews, stopped work events, and escalations that resulted from voice.

For each significant concern, record one of four outcomes: accepted with action, accepted with monitoring, rejected with technical reason, or unresolved. A high rate of unresolved items is a stronger warning than a low number of reports, because it shows the system consumes voice without learning from it.

5. Use meeting behavior as a leading indicator

Daily and weekly meetings reveal whether people can disagree before risk reaches the field.

A meeting in which the manager speaks for fifteen minutes and asks, "Any questions?" at the end is not a voice mechanism. It is a broadcast. Crew Resource Management in aviation shows why structured challenge matters, since hierarchy and time pressure can suppress information exactly when the team needs it most.

The operational test is simple. Count how many meetings include a challenge question, a risk assumption named aloud, and a documented answer. This does not require a new platform. It requires the supervisor to ask questions whose answers can change the job plan.

Use four prompts for a 30-day trial: what are we assuming, what could make this plan fail, who has seen this go wrong before, and what condition would make us stop. The quality of the answers will show whether the team is thinking or merely waiting for the meeting to end.

6. Connect speak-up data to incident investigation

Speak-up metrics become stronger when investigators check whether the warning existed before the event.

After a serious incident, teams often discover that someone had noticed the weak signal. The issue may have appeared as an informal comment, an unfinished corrective action, a maintenance note, or a concern that never left the shift. James Reason's work on latent failures helps leaders see that the silence around a risk can be part of the causal chain.

An incident investigation should include a voice timeline. Who saw what, when did they say it, who heard it, and what decision followed? This question avoids blaming the operator while still examining whether the organization had a chance to learn earlier.

When the same type of warning appears before multiple events, treat it as a system defect. The corrective action should not be another reminder to report. It should change the reporting path, leadership response, work design, or resource decision that made the warning ineffective.

7. Protect voice from becoming another compliance ritual

Speak-up programs fail when leaders convert human concern into a target that employees must feed.

The common failure is to set a quota for observations, reports, or safety conversations. Quotas create volume, although they can also create empty entries, copied language, and low-trust participation. Psychological safety is damaged when people feel that their voice is being harvested to satisfy a metric.

A stronger design uses minimum viable indicators with senior review. Track response time, voice quality, participation spread, dissent outcomes, meeting challenge behavior, and pre-incident warning history. That set is small enough for a C-level dashboard and specific enough to guide action.

Each month without a speak-up metric leaves leaders dependent on injury numbers, while the organization may already be receiving warnings that never become decisions.

Compliance reporting vs real speak-up intelligence

The difference between a reporting program and speak-up intelligence is whether the data changes leadership decisions.

Conclusion

Speak-up is not a soft culture topic when it shows whether risk information travels upward before an event forces everyone to listen.

Leaders who want real safety should put voice beside the indicators they already review, then ask whether employee concerns are changing work. For more conversations at the intersection of leadership and safety, visit Headline Podcast.