Psychological safety at work is one of the most cited ideas in modern management and one of the most consistently misunderstood. It gets read as a synonym for a nice atmosphere, which leads to team-building days, which changes nothing, which leads people to conclude the whole concept is soft.

Psychological safety at work is not about comfort. It is about whether saying the difficult thing is survivable.

What psychological safety at work actually means

The term comes from Amy Edmondson, whose 1999 study of 51 work teams defined it as a shared belief that the team is safe for interpersonal risk-taking.

Unpack “interpersonal risk” and it becomes concrete. It is the risk in admitting you do not understand. In saying the plan has a hole in it when the plan belongs to your boss. In flagging your own mistake early rather than hoping it resolves. In asking why we do it this way.

Every one of those carries a social cost — looking incompetent, looking negative, looking disloyal. Psychological safety at work is the shared expectation that the cost will not be charged.

Note that it is a property of a team, not a person and not a company. It is entirely normal for one department to have it and the department next door not to, because it is produced almost entirely by the behaviour of the person running the meeting.

The finding that explains why it matters

Edmondson’s early hospital research produced a result that looked wrong. The better-performing nursing teams appeared to make more medication errors than the weaker ones.

They were not making more errors. They were reporting them. The weaker teams were making the same mistakes and burying them, which meant nobody could learn from them and the same failures recurred indefinitely.

That is the mechanism in a sentence. Psychological safety at work does not stop things going wrong. It determines whether you find out.

Google reached the same place from a different direction. Its Project Aristotle study examined 180 teams looking for what made them effective, expecting to find the answer in composition — the right mix of skills and seniority. What it found was that team composition mattered much less than team norms, and psychological safety sat at the top of the list.

What it is not

  • Not niceness. Relentlessly pleasant teams are often the least safe, because disagreement has been socially priced out. Robust argument is a symptom of safety, not its absence.
  • Not lowered standards. Edmondson’s own framing pairs safety with accountability. High safety and low standards produces a comfortable team that achieves little; high standards without safety produces anxiety and concealment. You want both.
  • Not consensus. Nobody has to agree with you. They have to not punish you for saying it.
  • Not immunity. Poor performance still gets addressed. The difference is that it gets addressed rather than tolerated and gossiped about.

How to tell whether you have psychological safety at work

The problem with diagnosing psychological safety at work is that its absence looks like calm. Nobody objects. Nothing escalates. Meetings finish early.

Warning signs worth taking seriously:

  • The meeting after the meeting. The real conversation happens in the corridor, in a side channel, or at the pub. Everyone knew the objection. Nobody said it in the room.
  • Problems arriving fully formed. If issues surface as crises rather than as early concerns, they were visible to someone weeks earlier.
  • Nobody says “I don’t know”. In a room of ten people, that phrase should occur regularly. If it never does, everyone is performing certainty.
  • One voice dominates. Airtime concentrated in the most senior person is a reliable proxy.
  • Blank feedback forms. “No concerns” from an entire team is data, and it rarely means what it says.

How to build psychological safety at work

Edmondson’s own prescription is three moves, and they are all things a leader does rather than things an organisation announces.

  1. Frame the work as a learning problem, not an execution problem. “We have not done this before and we will get parts of it wrong” gives people permission to flag the parts.
  2. Acknowledge your own fallibility. “I may have missed something here — tell me” costs a leader almost nothing and changes what a room believes is allowed.
  3. Model curiosity. Ask a lot of questions. If the senior person asks genuine questions, questions become normal rather than evidence of not keeping up.

Three additions that work in practice:

  • Respond well the first time. Safety is built and destroyed at the moment someone takes the risk. One dismissive response to a challenge, in front of others, will undo six months of stated intent.
  • Speak last. If the person with authority states a view first, they have set the acceptable answer.
  • Separate the error from the person, publicly. Run a blameless review of something that went wrong and people will believe the next invitation to be honest.

Measuring psychological safety at work

There is a circularity here that trips people up: asking whether staff feel able to speak up requires them to feel able to speak up. In a team with low psychological safety at work, a named survey will return reassuring results, because telling the truth on it is exactly the interpersonal risk being measured.

Which is why genuine anonymity is not a nicety in this particular measurement. It is the only way to get a first reading from a team that does not yet trust the question.

It is why we treat anonymity as an engineering problem rather than a promise. Wotter’s anonymity assurance is designed so that no individual response can be traced back, which is what makes a first measurement of psychological safety at work worth taking seriously.

Measure at team level, not company level — averaging across a business hides exactly the variation you need to see. Watch the gap between teams rather than the absolute number, track it continuously rather than annually, and treat comment volume as a signal in itself. A team that suddenly starts writing more is usually a team that has started to believe someone reads it.

The uncomfortable version of this: if you introduce better listening and your numbers get worse, that is frequently progress. You have not created problems. You have started hearing about the ones that were already there — the same effect Edmondson found in those hospital wards, and a reasonable argument for measuring continuously rather than once a year.

Frequently asked questions

Common questions on what psychological safety at work is and how to build it. Wotter is an employee engagement platform — we measure this anonymously at team level, which is the only way to get an honest first reading. Anything else, drop us a line at hello@wotter.group.

A shared belief within a team that it is safe to take interpersonal risks — admitting you do not know, disagreeing with your boss, flagging your own mistake, questioning how things are done. The term comes from Amy Edmondson’s 1999 research.
No, and often the opposite. Relentlessly pleasant teams are frequently the least safe, because disagreement has been socially priced out. Robust argument is a symptom of safety rather than evidence against it.
No. Edmondson pairs safety with accountability. High safety and low standards produces a comfortable team that achieves little. High standards without safety produces anxiety and concealment. You want both at once.
Mostly the person running the meeting. It is a property of a team rather than a company, which is why one department can have it while the department next door does not.
Carefully — asking whether people feel able to speak up requires them to feel able to speak up. Anonymous responses at team level are the only reliable first reading, and the gap between teams matters more than the company average.