Psychological safety: what it is and how to build it

It is 11 a.m. on a Thursday. A sprint review is underway. The engineering lead pulls up the velocity chart and the numbers are bad — two weeks in a row below target. She asks the room: “Any thoughts on why this is happening?”

Silence. A few people study their laptops. Someone says “resource constraints.” Someone else nods. The meeting ends in twenty minutes and nothing changes.

Here is what nobody said: the new tooling decision made in week one added four hours of overhead per ticket. Three people on the team noticed this the first day. They said nothing, because the last time someone flagged a problem early in this organisation, the response was two hours of forensic questioning about why they had not caught it sooner. So they stayed quiet, finished the sprint short, and waited.

That meeting is not a communication problem. It is a psychological safety problem.

What psychological safety actually means

The term was brought into mainstream management research by Amy Edmondson (Harvard Business School) in a 1999 study on medical teams. She defined it as the shared belief held by members of a team that the team is safe for interpersonal risk-taking. The key word is “shared” — it is not a personality trait of a single brave person, it is a property of the group.

Interpersonal risks are things like: raising a concern that might sound naive, challenging a decision made by someone more senior, admitting you do not understand something, or reporting bad news early. In most workplaces, each of those actions carries a social cost. Psychological safety is the condition under which those costs are low enough that people act anyway.

Two things psychological safety is not, because they get conflated constantly.

First, it is not about being nice. A psychologically safe team can have very hard conversations. In fact, it has them more readily than a guarded team. The safety is precisely what makes directness possible — when you know the other person will not retaliate or dismiss you, you can say the difficult thing.

Second, it is not about lowering the bar. A team where nobody is held to account and poor work is accepted without comment is not psychologically safe — it is apathetic. Safety without standards produces politeness without progress.

The 2x2: safety and accountability together

The most useful model for thinking about this comes from Timothy Clark’s work on team dynamics, though similar matrices appear in Edmondson’s own writing. Plot any team on two axes: how psychologically safe it is, and how much accountability it carries. Four zones emerge.

Apathy Zone (low safety, low accountability). Nobody speaks, nobody is held to much either. Teams here are often described as “fine” in surveys. Output is mediocre and steady. The danger is invisible: there is no dramatic signal that something is wrong, so nothing gets fixed.

Anxiety Zone (low safety, high accountability). This is where demanding cultures without trust land. People are held tightly to results but live in fear of being blamed when things go wrong. The rational response is to hide problems until they are solvable — which is exactly the opposite of what a high-stakes environment needs. Errors compound in the dark.

Comfort Zone (high safety, low accountability). A genuinely nice place to work. Warm relationships, supportive managers. Also: deadlines are soft, underperformance goes unnamed, and the team slowly loses the habit of honest self-assessment. Comfort is easy to mistake for health.

Learning Zone (high safety, high accountability). People speak up because it is safe to do so. They are also expected to deliver, and when they do not, that is named. Bad news surfaces early enough to act on. Disagreement is treated as information. This is the zone where teams consistently outperform their circumstances.

What it looks like in practice — and what it does not

The sprint review at the opening of this piece was not a silence problem. The silence was downstream of something specific: a person was punished, visibly, for raising a concern early. Everyone else in the room updated their behaviour accordingly. That is the mechanism.

This is why psychological safety is not built through culture decks or town hall speeches about “speaking up.” It is built through repeated, specific behavioural signals, most of them small, nearly all of them from whoever holds power in the room.

Here is the pattern that matters: the leader’s response to bad news is the signal. Everything else is noise.

When a junior engineer tells you the integration is broken two days before the demo, you have a choice. You can become anxious, which expresses as urgency and blame (“why didn’t you catch this sooner? What were you doing all week?”). Or you can respond with curiosity: “Okay. Walk me through what you found. What are our options from here?” The first response takes five seconds and sets back psychological safety by weeks. The second takes the same five seconds and advances it.

The junior engineer will tell the next ten people on the team what happened in that conversation. That is how norms propagate.

Four things leaders actually do

1. Admit your own mistakes first, in public. Not performatively — specifically. “I called the timeline wrong on the partner integration. I underestimated the dependency chain. Here is what I am doing differently.” When the person with the most to lose from admitting error does it openly, it becomes safe for everyone else to do the same. If you only admit failures after they are obvious to everyone, you are not modelling safety — you are doing damage control.

2. Invite dissent explicitly. “What is the strongest case against this plan?” is a different question from “any concerns?” One signals that you are genuinely looking for the hole; the other signals that you are looking for assent. Red-teaming — asking one person to argue against a decision — is a structured version of this. It removes the social cost of objection by making it a role rather than a personal stance.

3. Reward the messenger. The strongest signal you can send is what you do the first time someone brings you a problem that has no solution yet, or one that reflects badly on the team. If that person leaves the conversation feeling glad they told you, word gets around. If they leave feeling interrogated or implicitly blamed, word also gets around. The reward does not have to be large. “I am really glad you flagged this early” and then moving to problem-solving mode is enough.

4. Separate the post-mortem from the performance review. When a project fails, the most valuable thing is understanding what went wrong so the team can learn from it. That requires people to be honest about their role in the failure. They will not be honest if the debrief feels like evidence-gathering for a later evaluation. Keep after-action reviews explicitly separated from performance consequences, and say so at the start of each one.

The accountability half of the equation

A point that gets lost: psychological safety does not mean every idea is treated as valid or every deadline as optional.

The manager who never gives critical feedback, who lets poor work slide to preserve warmth, is not creating safety. They are creating a team that does not know where it stands, cannot improve, and eventually loses trust in the manager’s candour. That is its own kind of unsafe.

High accountability in a high-safety environment sounds like: “I am giving you this feedback because I think you can do better, and because I want you to know what I am seeing.” It is specific, it is timely, and it assumes competence and good faith on the other side. What it is not: vague disappointment delivered months after the fact, or criticism deployed as a status move in a group setting.

The pair matters. Safety without accountability is comfort. Accountability without safety is anxiety. The Learning Zone is not a moderate amount of each — it is a genuine high on both dimensions simultaneously.

A note on seniority

Psychological safety is not symmetrically distributed across a hierarchy, and it would be naive to pretend otherwise. The most junior person in the room has the most to lose from speaking, and the least power to change the dynamic if it goes badly. The most senior person has the least to lose and the most ability to set the norm.

This means the work of building psychological safety belongs disproportionately to whoever is most senior. It cannot be delegated to HR, it cannot be outsourced to a team offsite, and it does not emerge spontaneously from good intentions. It requires the person with the most formal power to act in ways that deliberately reduce the cost of speaking for those with the least.

The sprint review at the start of this post could have gone differently. Not if a junior team member had decided to be braver. If the engineering lead had, in the previous month, responded to one piece of bad news with visible curiosity instead of audible stress — the room would have been different.

That is the work. It is small, it is repeated, and it compounds exactly the same way the silence does.