Most managers who have a team that never challenges them think they have a respectful, well-run team. Some of them do. More often, what they have is a team that has learned it is not safe to disagree — and has quietly stopped trying. The meeting where everyone nods, the project where problems surface only after they’ve become expensive, the brilliant hire who leaves after eighteen months because she never felt she could say what she actually thought: these are not communication problems or culture problems in the abstract. They are the predictable output of a team without psychological safety.
This is a guide for managers who want to understand what psychological safety actually is, what the evidence says about why it matters, and — most importantly — what practical steps build it.
What psychological safety is (and is not)
The term comes from Amy Edmondson, a professor at Harvard Business School, who first identified the concept while studying medical teams in the 1990s. She noticed something counterintuitive: the best-performing hospital units reported more errors, not fewer. Her explanation — confirmed by two decades of subsequent research — was that the high-performing teams had created conditions where staff felt safe flagging problems early. The lower-performing units had fewer reported errors because their members had learned that reporting mistakes was punished. The errors were still happening. Nobody was saying so.
Edmondson defines psychological safety as a shared belief that the team is safe for interpersonal risk-taking — specifically, that you will not be punished, humiliated, or marginalised for speaking up, asking questions, pointing out problems, or disagreeing with the person above you.
This is not the same as comfort. Psychologically safe teams are not uniformly agreeable or conflict-free. They often have sharp disagreements, because the disagreements are out in the open rather than muttered over coffee after the meeting. Psychological safety is the condition that makes candour possible — it is not a ban on challenge; it is the thing that allows challenge to happen productively.
It is also not a personality trait of the team members. Whether a team feels safe to speak up depends overwhelmingly on the behaviour of the manager. Edmondson’s research is unambiguous on this point: the primary determinant of a team’s psychological safety is how the leader responds when things go wrong or when someone says something inconvenient.
What the evidence says
The case for psychological safety is not theoretical. The most influential piece of corporate research on the subject is Google’s Project Aristotle — a multi-year study of 180 internal teams, published in 2016 — which set out to answer a deceptively simple question: what makes a Google team effective?
The researchers had expected the answer to involve the composition of the team — the right mix of skills, seniority, and personalities. What they found, after analysing hundreds of variables, was that team composition barely mattered. The dominant factor — the one that predicted team performance more than any other single variable — was psychological safety. Teams where members felt safe to take interpersonal risks consistently outperformed teams that did not, regardless of individual talent.
Google’s findings echoed and reinforced what Edmondson had been publishing for years. High psychological safety predicts: greater willingness to report problems early; faster learning from failures; higher-quality decision-making (because more information reaches the people who need it); stronger engagement and retention; and, in knowledge-work contexts, meaningfully better performance outcomes.
The mechanism is not mysterious. Decisions are only as good as the information that goes into them. If team members withhold bad news, hedge their real views, or avoid raising concerns because previous experience has taught them that doing so is costly — then every decision made at the top of the team is made on incomplete information. Psychological safety is, at its core, an information problem disguised as a culture problem.
What destroys it
Before talking about what builds psychological safety, it is worth being precise about what damages it — because the culprits are often not dramatic. Psychological safety does not usually die because a manager shouts at someone in a meeting. It erodes through smaller, consistent signals.
The dismissive response. “We’ve tried that before.” “That’s not really relevant here.” “Let’s park that and keep moving.” These responses are rarely intended as shutdowns, but they function as one. The person who suggested the idea noted the response. So did everyone else in the room. The next time they have a thought they are uncertain about, they will do the calculation faster.
The punished mistake. If a team member raises a problem, admits an error, or flags a risk and the response is blame, public criticism, or a visible deterioration in how they are treated — the lesson is learned by everyone who witnesses it, not just the person involved. The chilling effect of a public dressing-down outlasts the incident itself by months.
The performative question. Asking “does anyone have any concerns?” in a tone that makes clear the expected answer is “no” does not build safety. It models that questions are formalities rather than invitations.
Favouritism in response. If certain team members’ ideas are consistently received more warmly — because they are more senior, more confident, or simply more aligned with the manager’s existing view — the team will figure out very quickly who is actually worth listening to. Everyone else adjusts their behaviour accordingly.
Silence after good challenges. Psychological safety is not destroyed only by negative responses. If someone raises a difficult but valid point and the manager neither acknowledges it nor returns to it, the signal is the same: speaking up leads nowhere.
Four things managers can do
1. Set explicit expectations about candour
The manager who says “I want you to tell me when things are going wrong, even when — especially when — it’s uncomfortable” and then behaves consistently with that statement is doing something genuinely valuable. The statement alone means nothing; the consistency is what counts. But making the expectation explicit is still useful, because it signals intention before behaviour has had time to demonstrate it.
A specific variation that works: when setting up a project or entering a new phase of work, name the kinds of information you want surfaced. “If we’re at risk of missing a milestone, I want to know early — even if the news is bad and especially if you think I won’t like it.” This is not a general invitation to complain; it is a specific statement about what information you need and a reassurance that providing it will not backfire.
2. Model fallibility visibly
Edmondson’s research identifies this as one of the most powerful single behaviours available to a manager: being visibly fallible in front of the team. Admitting uncertainty. Saying “I don’t know” rather than speculating confidently. Describing a time when your judgement was wrong. Asking the team for their view on a decision you have not yet made.
The logic here is not that appearing less competent somehow inspires confidence. It is that managers who project infallibility — whether deliberately or because showing uncertainty feels professionally risky — create a norm in which everyone else must also project infallibility. The team learns that uncertainty is not acceptable. The predictable result: problems are concealed until they are too visible to hide.
When a manager demonstrates that they can be wrong, ask for help, and survive it professionally, they are setting a different norm. The implicit message is: in this team, not knowing something is a problem we can state out loud rather than something to be managed quietly.
3. Respond to bad news as information, not as accusation
This is the most important lever and the hardest to sustain under pressure. When a team member tells you something has gone wrong — a client is unhappy, a deadline is at risk, an assumption turned out to be incorrect — your first response determines how frequently you will hear things like that in future.
The response that builds psychological safety is curious and forward-looking: “Thanks for flagging this early. Tell me what’s happened and what you need.” The response that kills it is interrogative and backward-looking: “How did this get to this point? Why wasn’t I told sooner?”
The second response is natural. It is also almost always counterproductive. The message it sends is that bringing bad news has a cost, and the team will work to avoid paying that cost in future — which means you will hear about problems later, when they are larger and harder to resolve.
This does not mean failures should have no consequences, or that standards do not matter. It means the conversation about what went wrong and how to prevent it is more valuable than the conversation about who is to blame. The first improves the team; the second mostly just ensures that next time, the team hides the failure longer.
4. Actively solicit dissent
High psychological safety does not mean the manager waits for people to volunteer challenge. It means actively creating conditions where challenge is expected, invited, and rewarded when it happens.
Practically, this might mean:
- Going round the table explicitly: “Before we close on this, I want to hear from anyone who has a reservation we haven’t fully worked through.”
- Assigning someone to argue against a proposal — not as a performative exercise, but because it surfaces the strongest counterarguments before the decision is made, not after.
- Following up with the person who stayed quiet: “You didn’t say much in the meeting — I’d value your take when you’ve had a chance to think about it.”
- Publicly thanking someone for raising a difficult point, particularly if the point was correct and you didn’t want to hear it: “That was the right thing to flag. I’m glad you did.”
The last behaviour is disproportionately powerful. When the team sees that speaking up is rewarded rather than merely tolerated, the risk calculation changes for everyone watching.
How to know if it’s working
Psychological safety is not directly measurable in the way productivity or headcount is, but there are observable proxies that a manager can track.
The most reliable early signal is the quality and frequency of bad news reaching you. If you are only hearing about problems once they are too large to ignore, that is a signal. If team members flag risks early — sometimes before you would have thought to ask — that is the opposite signal. The question “does my team tell me things I would rather not hear?” is a more useful diagnostic than any engagement survey.
Other signals: the ratio of genuine questions to rhetorical affirmations in meetings; whether quieter team members speak at all, and whether what they say influences anything; whether mistakes are discussed openly or quietly absorbed; whether people ask for help or only surface problems they have already solved.
Edmondson developed a seven-item survey that measures team psychological safety directly, and it is widely used in research and organisational settings. For most working managers, however, the survey is secondary to honest observation of these behavioural markers. Numbers on a questionnaire can be gamed, consciously or unconsciously, in teams where people have learned that honest answers are themselves a risk.
The manager’s own safety
One dimension that is often overlooked: psychological safety is not only about the relationship between manager and team. It also depends on the safety the manager experiences in their own reporting line.
A manager who is under pressure to project confidence upwards — who cannot admit uncertainty to their own boss, who is penalised for surfacing problems rather than solutions, who must always appear in control — will, with very few exceptions, recreate those dynamics in the team below them. This is not a character failure. It is the predictable transmission of an incentive structure.
If you want to build psychological safety in your team, it is worth examining honestly whether your own manager creates the conditions that make that possible — and, where they do not, what you can reasonably model within the bounds of your own authority. You cannot give your team something you have not been given. You can, however, create a small circle of relative safety within a wider culture that lacks it — and that matters more to the people on your team than any external programme.
Related reading
If you manage a team and want a structured CPD pathway for building safer, higher-performing team cultures, SkillsCircle — built by Accipio — offers accredited management development programmes that go well beyond the checklist. Aicura, Accipio’s AI coaching tool, can also support managers in practising the conversations that psychological safety depends on.