Psychological safety at work: what it is, and what it is not

Psychological safety is the single most important factor in how well a team performs, according to Google's research into its own teams. This article covers what the term actually is, what it is often confused with, why it is harder than it sounds, and four leader behaviours that build it.

By Caroline Spännar · Reading time approx. 9 minutes

The single most important factor in how a team performs

When Google wanted to understand what separates its best teams from the rest, it analysed around 180 of its own teams in Project Aristotle. They tested level of education, personality types, leader profile and a range of other factors, without any of them explaining the differences reliably. The factor that correlated most strongly with how well teams performed was psychological safety.

The effect does not stop at performance. Teams with high psychological safety learn faster, because people dare to try, ask and admit mistakes in time. They also sustain engagement better, because it is easier to take part in a group where you dare to say what you think. Amy Edmondson's original research shows exactly that chain: safety leads the team to learn from its experiences, and that learning is in turn what drives results.

What psychological safety actually is

The term comes from Amy Edmondson, professor at Harvard Business School. It grew out of her early studies of hospital teams and was formulated in a groundbreaking paper in 1999, where she gave it its now-standard definition:

Psychological safety is a shared belief held by members of a team that the team is safe for interpersonal risk-taking.

The heart of it lies in the words interpersonal risk. It is not about physical or legal safety, but about the small everyday risk of saying something that could make you look bad: asking a question that might seem uninformed, admitting a mistake, challenging your manager in front of the group, or putting forward an idea that may be rejected.

In a team with high safety, people take those risks as a matter of course. In a team without it, they avoid them, and the group loses precisely the information it needs most.

One of Edmondson's early discoveries in healthcare illustrates this well. When she measured error reporting across hospital units, the best-functioning teams reported more errors than the others, not fewer. It was not because they made more mistakes. On those units the staff dared to talk about the errors, while the units with weaker leadership hid them and lost the chance to learn.

What it is often confused with

The most expensive misconception is that safety means niceness. Then the measures taken are wrong, and the results fail to appear.

Safety is not consensus. Edmondson is clear that teams with high psychological safety often have more open disagreement than others, because people actually dare to say what they think. The difference is that the conflict is about the issue, not about the person.

It does not lower the bar. In her later research Edmondson describes how the strongest teams combine high safety with high expectations, a combination she calls the learning zone. High safety with low demands produces comfort without results, and low demands with low safety produce indifference. It is the interplay between safety and demands that makes the difference, not safety on its own.

It does not mean freedom from accountability. Being able to say "I got it wrong" does not remove responsibility for the mistake. If anything, accountability is greater in safe teams, because people dare to own their mistakes openly rather than hide them.

And it is not HR's job alone. HR can train, measure and hold the structure, but safety is built or broken in the daily contact between manager and employee. Every manager owns the climate in their own team.

That last point is actually one of the most encouraging things in the research. A team can be psychologically safe relatively independently of the organisation around it. A manager can therefore start building safety in their own group without first waiting for the rest of the system, unlike much else in modern organisations where you are stuck in structures you cannot control. It helps, of course, if the leadership above also has high safety, but it is not a precondition for getting started.

Why it is harder than it sounds

There is a simple explanation for why psychological safety is hard to build. Most of us want to appear competent, and so we tend to avoid talking about our mistakes or asking questions that might reveal what we do not know. We also want to be seen as pleasant and easy to work with, and so we hold back from questioning the group's opinions. The behaviour is called impression management, and it is deeply human. We want to fit in. The problem arises when the wish to make a good impression outweighs the task.

On top of that, it is easy to know how you ought to behave, but hard to step out of autopilot when you are challenged. Psychological safety is built through awareness in the moment: recognising your defensive reflexes, holding back the first reaction, and choosing a constructive response rather than a quick one.

Four leader behaviours the research points to

Edmondson's work and a growing body of follow-up studies have distilled a number of concrete behaviours that affect safety directly. Four of them recur.

1. Acknowledge what you do not know

The strongest signal a manager can send is to say "I do not know" or "I was wrong" in front of their team. It normalises not having all the answers and frames the task as something the team solves together, rather than something each person should manage alone.

Concrete behaviour: say it explicitly at least once a week in a group meeting, for example "I do not know" or "I am changing my mind about this after hearing you".

2. Actively invite dissent

"What do you think?" rarely does the job, because it is usually only those who already feel safe who answer. Targeted invitations work better: "Anna, you have worked with this before, what do you see that we are missing?" Or turn it around: "What is the strongest argument against the proposal?"

Concrete behaviour: set aside a couple of minutes in every decision meeting to look for the dissenting perspective, and thank the person who raises it even when the decision lands elsewhere.

3. Treat mistakes as information

When someone flags a problem, the manager's reaction in that very moment decides what happens next time. If the person who speaks up is met with irritation or a "why did you not catch this earlier?", the team learns to stay quiet.

Edmondson distinguishes between mistakes that could have been avoided, mistakes that stem from complexity, and mistakes that belong to exploratory work, and the three call for different responses. Safety falls when a manager treats them all as the first kind.

Concrete behaviour: begin with "thank you for saying that, what do we need to understand?" before you ask why.

4. Talk about your own uncertain moments

The hardest, and at the same time the most powerful. When a manager describes a time they were uncertain or failed, and what they learned, the threshold drops for everyone else in the room.

Concrete behaviour: share one such specific situation once a quarter, ideally where the whole team hears it.

How you measure psychological safety

It can be measured, and if it is not measured it is rarely prioritised. Edmondson has developed a measurement tool based on seven criteria that has been validated in thousands of studies. Short versions usually contain statements along these lines:

• If I make a mistake in this team, it is often held against me.
• It is safe to take a risk in this team.
• In this team, it is easy to ask for help.
• My perspectives are valued in this team.

The answers are given on a scale. An aggregated score per team is meaningful, but an aggregated score for the whole company is not, because safety varies dramatically between teams in the same organisation. It is often in that variation that the most important signals lie.

So put psychological safety into recurring pulse surveys rather than the annual employee survey. The effect of an intervention appears within two to three months, and if you wait a whole year you lose the link between measure and outcome.

Three common pitfalls when organisations start

Believing that a lecture is enough. Psychological safety is a muscle, not a piece of knowledge. A one-off lecture for 200 managers very rarely changes everyday behaviour. What works is recurring training in concrete tools and reflection in smaller groups: 8 to 30 people, 60 to 90 minutes, every other week over a few months.

The top not modelling it. If leaders never show vulnerability themselves, never admit mistakes, never flag uncertainty, then all the training of employees further down the organisation is only half effective. Edmondson's studies of hospitals showed that the psychological safety of medical teams was mainly governed by how senior doctors behaved. The same principle applies in companies.

Confusing it with job satisfaction. A team can be perfectly happy and still have low psychological safety, if people never challenge each other. Another can have high safety and at the same time be completely exhausted, if the demands are unreasonable. Job satisfaction and safety measure different things.

How you start

The most common mistake is to try to roll out psychological safety broadly across the whole organisation at once. That works poorly. What works is starting narrow, with one or two teams where:

1. the manager wants to rather than has to,
2. the need is clear (falling results, high sickness absence, quality problems),
3. and you can measure before and after.

Run a structured programme over three to four months that alternates knowledge, practice and reflection, and measure safety before, after and some time after the end. When the figures come back, you can decide whether the model should be scaled up. If they show no progress, you have still learned something, cheaply and without having burnt the budget on a broad rollout.

This topic connects to several others: how a pilot programme is built in practice, what the rising level of sickness absence costs, and what the employer's responsibility under the Swedish work-environment regulation AFS 2023:2 involves. We cover those in separate articles here in Knowledge.

That is roughly the model we have built our service Course: Psychological safety around, a structured programme based on Edmondson's research. If this topic is relevant for you, a 20 minute intro call is often the simplest way to see whether it fits your situation.

Summary

Psychological safety is a measurable quality of teams that research links to performance more strongly than any other single factor. It is built above all in a manager's everyday behaviour, in how they handle what they do not know, seek dissent, receive mistakes and show their own uncertainty.

It requires no large apparatus, but it does require that leaders want to and are given the tools to start. And given what mental ill health costs employers today, it is arguably one of the most underestimated investments a company can make.