Psychological Safety at Work: What It Is and How to Build It
Psychological safety is a team's shared belief that it is safe to speak up, take risks, and admit mistakes. Here is what the research shows and what managers can do about it.
Psychological safety is the single strongest predictor of team performance identified across decades of organizational research. It is also one of the most misunderstood concepts in management. This article defines it precisely, shows what the data says it produces, and gives managers a practical framework to build it -- including the part most training programs skip.
What Is Psychological Safety at Work?
Psychological safety is a shared belief held by members of a team that the team is safe for interpersonal risk taking. That definition comes from Amy C. Edmondson, Professor of Leadership and Management at Harvard Business School, who first articulated it in a 1999 paper published in Administrative Science Quarterly.
The word "shared" matters. Psychological safety is not a trait of confident individuals. It is a property of the team environment. A highly confident person can feel psychologically unsafe in the wrong team. A naturally reserved person can speak up freely in a team where the conditions are right.
The word "interpersonal risk" matters too. Psychological safety is not about comfort or the absence of accountability. It is about whether team members believe they can raise a concern, admit a mistake, or challenge a decision without suffering social punishment for doing so.
Edmondson distinguishes this clearly from trust, which is typically a property of a one-to-one relationship. Psychological safety is collective. It is the climate of the group.
Where the Research Comes From
Edmondson's 1999 study examined 51 work teams in a manufacturing company. It found that psychological safety was strongly associated with learning behavior -- the willingness to experiment, ask for help, and talk openly about errors. Her subsequent book, "The Fearless Organization" (2018), synthesised those findings across healthcare, technology, manufacturing, and professional services.
The defining large-scale test of the concept came from Google. In 2012, Google launched Project Aristotle: a study of 180 internal teams (115 engineering, 65 sales) designed to identify what separates high-performing groups from average ones. The researchers analyzed data across personality traits, team composition, communication patterns, and management styles.
Psychological safety emerged as the most important variable. Individuals on teams with higher psychological safety were less likely to leave Google, more likely to contribute diverse ideas, more likely to bring in more revenue, and rated as effective twice as often by senior leaders.
Google published these findings through its re:Work platform. The conclusion was unambiguous: no other factor -- not team composition, seniority mix, or management style -- predicted team performance as consistently as whether team members felt safe to speak.
What Low Psychological Safety Costs
The cost of low psychological safety is not abstract. The American Psychological Association's 2024 Work in America survey -- conducted by The Harris Poll across 2,027 employed adults -- found that 61% of workers in low-safety environments typically feel tense or stressed in a standard workday. Among workers in high-safety environments, that figure falls to 27%.
The retention gap is equally sharp. The same survey found that workers experiencing lower psychological safety have more than twice the intent to leave their job within the next year: 41% compared to 19% in higher-safety environments.
BCG's January 2024 research put a number on the attrition difference directly. When psychological safety is high, only 3% of employees plan to quit. When it is low, that figure rises to 12%. BCG also found that psychological safety makes employees 2.1 times more motivated and 3.3 times more likely to feel they can reach their full potential at work.
The APA survey also found that 59% of workers believe their employer thinks their work environment is significantly healthier than it actually is. That gap between management perception and employee experience is the operational consequence of teams where people do not speak up.
Edmondson's Four Zones: Safety Is Not Enough on Its Own
A common misreading of psychological safety is that more is always better. Edmondson's framework in "The Fearless Organization" is more precise than that. She maps team environments across two dimensions: psychological safety on one axis, accountability on the other.
This produces four zones:
Apathy Zone -- Low safety, low accountability. People do the minimum to avoid consequences. Initiative disappears.
Comfort Zone -- High safety, low accountability. People feel free to speak but lack the drive to perform. Teams become complacent.
Anxiety Zone -- Low safety, high accountability. People fear punishment for errors. Problems get hidden rather than raised. This is where catastrophic failures incubate.
Learning Zone -- High safety, high accountability. The optimal state. Teams take intelligent risks, share information openly, maintain high standards, and deliver results.
The Learning Zone requires both elements. A manager who builds psychological safety while dropping expectations has moved their team into the Comfort Zone. The goal is not to make people feel good. The goal is to make it safe to do excellent work and admit when it is not going well.
How Managers Build Psychological Safety
Psychological safety is not built through policy. It is built through repeated, observable manager behavior -- specifically through how managers respond when team members take interpersonal risks.
Edmondson identifies three core leader behaviors in "The Fearless Organization": setting the stage by framing work as a learning problem, inviting participation by acknowledging personal fallibility, and responding productively when people do speak up. All three behaviors signal to the team what the actual rules are.
1. Frame work as a learning challenge, not a performance audit
When managers communicate that the goal is to learn and improve, they lower the cost of admitting uncertainty. When every interaction is evaluated against a performance standard, people optimize for appearing competent rather than being honest.
The practical version: open team conversations with what you do not yet know. Ask questions before making declarations. Treat error analysis as diagnostic, not disciplinary.
2. Acknowledge your own limitations explicitly
This is the behavior most managers have never practiced and most training programs never teach. Research published in Scientific Reports in 2025 found that when leaders model speaking up -- including voicing uncertainty and concern -- followers do the same. The mechanism is psychological safety: leader behavior signals what is acceptable and what is not.
Acknowledging a limit sounds like: "I got that wrong last quarter" or "I am genuinely uncertain about this -- I need your view." These are not signs of weakness in an organization with high safety norms. They are load-bearing signals that it is acceptable to be honest.
3. Respond visibly well when someone takes a risk
This is the highest-leverage single behavior. When a team member raises a concern, admits a mistake, or challenges a decision, the manager's response in that moment sets the terms for every future interaction.
A productive response acknowledges the contribution, separates the information from any judgement about the person, and acts on what was raised. An unproductive response -- dismissiveness, deflection, or silence -- sends the message that interpersonal risk is punished. Once that signal is received, teams stop sending the signal.
4. Explicitly invite dissent
Most people default to agreement in meetings. Managers who rely on the absence of objection as a form of consensus are reading silence incorrectly. The practical corrective is to ask specifically for challenge: "What am I missing?" or "Who sees this differently?" These questions normalize a different norm.
This gap -- surfaced in Wiley Workplace Intelligence's 2023 research -- illustrates why top-down perception of team health is unreliable. Executives experience their own meetings as high-safety. The people below them often experience theirs differently.
The Uncomfortable Truth About Building Psychological Safety
Managers cannot build psychological safety by announcing it. They build it by modeling the behaviors themselves -- and that requires being willing to be visibly uncertain, visibly wrong, and visibly open to challenge.
This is a harder ask than it sounds. Most managers have been selected and promoted for appearing competent. The behaviors that build psychological safety -- admitting errors, inviting disagreement, sitting with uncertainty -- run against the performance norms that got them to the role.
It is also a skill that most managers have never deliberately practiced. Knowing that you should model vulnerability is one thing. Being able to do it in a live team meeting -- with a pressing deadline, a difficult relationship, or a credibility concern in the room -- is another. The knowledge does not automatically produce the behavior.
This is where the training approach matters. A manager who has had a chance to rehearse these conversations -- to practice saying "I was wrong about that" in a low-stakes environment, to try a different approach when their first attempt lands badly, to build the conversational fluency before the real moment arrives -- has a better chance of doing it when it counts.
Ambr AI builds bespoke voice simulations so managers can practice the conversations that build psychological safety before they matter most.
See how it worksIn a pilot program with Skyscanner, managers who completed 12 weeks of regular AI-based conversation practice with Ambr AI showed measurable change. 78% reported feeling more comfortable navigating difficult conversations. Engagement across the cohort reached 92% -- well above typical rates for behavioral development programs.
A Practical Framework for L&D Teams
Organizations that take psychological safety seriously as a business priority need a different measurement approach. Not "did we run a psychological safety workshop?" but "have our managers demonstrably changed how they respond when people speak up?"
That distinction -- between training delivery and behavior change -- determines whether the investment produces results.
A practical approach has three steps:
Step 1: Measure the baseline. Survey team members on whether they feel comfortable raising concerns, admitting mistakes, and challenging decisions with their manager. Do not rely on manager self-report. The APA data shows the perception gap is substantial and systematic.
Step 2: Build practice, not awareness. Awareness of psychological safety does not change manager behavior under pressure. Managers need repeated, specific practice of the actual behaviors -- acknowledging errors, asking for challenge, responding well when someone takes a risk. This requires a different learning design than a workshop.
Step 3: Track behavioral signals, not training completion. The relevant indicators are whether speaking-up frequency in team settings has changed, whether manager self-disclosure behavior has shifted, and whether team members report a measurable change in how concerns are received. Completion data tells you the training happened. Behavioral data tells you whether it worked.
What is psychological safety at work, and where does the concept come from?
Psychological safety is a shared belief held by members of a team that the team is safe for interpersonal risk taking. The term was defined by Amy C. Edmondson, Professor of Leadership and Management at Harvard Business School, in a 1999 paper published in Administrative Science Quarterly. It describes a property of the team climate -- not an individual trait -- and is associated with learning behavior, error reporting, and sustained high performance.
How does psychological safety differ from trust or comfort at work?
Psychological safety is a collective property of a team rather than a bilateral relationship between two people. Trust is typically a one-to-one dynamic. Psychological safety is the shared climate of the group. It also differs from comfort: Edmondson's framework explicitly maps a Comfort Zone where safety is high but accountability is low. The goal is not comfort -- it is a combination of high safety and high accountability that she calls the Learning Zone.
What did Google's Project Aristotle find about psychological safety?
Google's Project Aristotle -- a large-scale internal study launched in 2012 and published via Google's re:Work platform -- found that psychological safety was the most important factor distinguishing high-performing teams from average ones. Individuals on high-safety teams were less likely to leave, more likely to contribute diverse ideas, more likely to generate revenue, and rated as effective by senior leaders twice as often as those on low-safety teams.
What does low psychological safety cost an organization?
The APA's 2024 Work in America survey found that 61% of workers in low-safety environments feel stressed in a typical workday, compared to 27% in high-safety environments. BCG's 2024 research found that attrition intent rises from 3% to 12% when psychological safety is low. Workers in low-safety environments are also less motivated, less productive, and more likely to believe their employer fundamentally misunderstands the quality of their work environment.
What specific behaviors can managers use to build psychological safety on their team?
The four highest-impact behaviors are: framing work as a learning challenge rather than a performance audit; explicitly acknowledging personal limitations and mistakes; responding visibly well when team members take interpersonal risks; and actively inviting dissent rather than treating silence as agreement. Each behavior sends a signal about what the real norms are. Psychological safety is built through the accumulation of those signals over time, not through a single policy or announcement.
Why do managers struggle to build psychological safety even when they understand it?
Understanding the concept does not produce the behavior. Managers are typically selected and promoted for appearing competent and decisive. The behaviors that build psychological safety -- admitting errors, inviting challenge, sitting with uncertainty openly -- run against those learned performance norms. Modelling vulnerability in a live team setting is a learnable skill, but it requires deliberate practice in conditions close enough to the real situation to build transferable fluency. Most managers have never had that practice.
What is the difference between psychological safety and psychological comfort?
Psychological safety means believing it is safe to take interpersonal risks -- to raise concerns, challenge decisions, and admit mistakes. Psychological comfort is the absence of challenge or discomfort. The two are not the same. Edmondson's research identifies a Comfort Zone where safety is high but accountability is low, and teams become complacent. High psychological safety combined with high accountability -- the Learning Zone -- produces performance. Comfort alone does not.
How should L&D teams measure whether psychological safety is improving?
Measure behavioral signals rather than training attendance. Relevant indicators include how frequently team members raise concerns or challenge decisions in meetings, whether managers demonstrate self-disclosure and error acknowledgment in observed settings, and what team members report when surveyed directly. The APA's 2024 data shows a substantial gap between manager perception and team experience, which means manager self-report is an unreliable measure. Ask the people below.
Ambr AI builds bespoke voice-based conversation simulations for enterprise workplace training, built around the specific scenarios, language, and culture of each client.
Sylvie Waltus
Marketing Manager
Continue reading
Related reading
Mar 3, 2026
Why First-Time Manager Training Fails (and What to Do Instead)
82% of new managers receive no formal training before leading a team. Here's why traditional programs don't work — and what the evidence says actually does.
Feb 19, 2026
The Hidden Cost of Avoiding Difficult Conversations at Work
Avoiding difficult conversations costs US organizations $359 billion annually. Here is what the data shows about conversation avoidance, why managers avoid it, and what actually moves the needle.
Jan 21, 2026
How Poor Management Causes Employee Turnover — and What to Do About It
Over half of employees who leave say their manager could have prevented it. Here is what the Gallup data shows about manager-driven attrition and the highest-ROI interventions to stop it.