“Good failures happen in new territory”

When people have permission to be open, they can speak honestly about what is going wrong. This is the basic idea behind the term "psychological safety", coined by Harvard professor Amy Edmondson. In the interview led on last year's Peter Drucker Forum, the management thinker talks about how psychological safety enables learning and what influence an uncertain environment has.

Foto: Lea Sonderegger
Foto: Lea Sonderegger

Two types of safety

Companies are currently dealing with a wide variety of crisis hotspots and are therefore exposed to a high degree of uncertainty. How does this affect the psychological safety of people in organisations?

There are two types of safety. One is interpersonal safety: I am not afraid to speak up because I don't have to fear that others will judge me. The other form is physical or literal safety from a variety of threats. There are things we are right to consider unsafe or worrisome – be it a virus, economic downturn, or a tragic war. The key is that real threats to safety are best managed when people can talk about them openly, so as to think together about what to do about them. This means that there is both productive and counterproductive fear. It’s productive to be afraid of a contagious virus; it’s counterproductive to be afraid of what others might think of your novel idea or dissenting viewpoint. When we don't make this distinction, we run the risk that people will not speak up or take action – because they’re inhibited by counterproductive fear – and thereby fail to do things that could help them out of the situation.

Psychological safety has tended to decrease in the face of the many new challenges in the external environment.

So, has psychological safety currently increased or decreased?

It has tended to decrease in the face of the many new challenges in the external environment. But that doesn't have to be the case. When we humans are in dangerous situations, the amygdala in the brain intervenes: it makes us withdraw. We hide in a way, because we don't want to make the situation worse by making a mistake. This behaviour lulls us into a false sense of security and reduces psychological safety.

Global Peter Drucker Forum 2023

Amy Edmondson will be speaker at this year's Global Peter Drucker Forum that takes places Nov 30 to Dec 1 in Vienna.

Thursday, 30 November 2023

12:00 - 13:00 CET

PANEL 3: Making the Most of Failure



“Good judgment comes from experience,” the old quote goes, “but unfortunately, experience comes from bad judgment.” Why do some people, organizations, sectors never recover from mistakes, while others learn and grow?


Julia Kirby Senior Editor, Harvard University Press


Amy Edmondson Professor of Leadership & Management, Harvard Business School

Daniel Pink Best-selling author

Niren Chaudhary Chairman of the Board, Panera Brands

Julie Teigland Member of the Global Executive & Area Managing Partner for EMEIA, EY

More about the Global Peter Drucker Forum 2023

But psychological safety can be increased if uncertain or threatening situations are handled properly: specifically, external threats must be framed as opportunities to collaborate, learn quickly, and innovate to find solutions.

Talking can help

When real difficulties arise in companies, do you think just talking about them helps?

Yes, if we talk about the reality in a thoughtful way. For example, if there are layoffs, that will obviously affect the psychological safety of those who stay in the company. But good leaders will say things like, "I know this situation will lead people to want to hold back and not express risky ideas, not ask for help, not admit mistakes. But in times like these, we need more of that, not less." So, they name the risk and reframe what is wanted. They reframe the default belief from people thinking it’s best to hide to perhaps thinking it’s best to contribute.

You have been researching psychological safety for many years. How well has this concept been implemented in companies?

The good news is that there is variance across work groups in every company I’ve studied. But that is also the bad news at the same time. What does that mean? I have not yet found a company in which there are not, at least in isolated cases, some great, fearless teams with climates for speaking up openly in difficult situations. But in the same company, elsewhere, you find teams with very low psychological safety.

How do you explain that?

An implication of this is that psychological safety is not determined by top management alone. Top management matters, of course. The example senior executives show has a very tangible effect on the culture, but that effect is filtered through middle managers - team leaders, branch managers, project managers. These are the people whose direct leadership qualities have the greatest impact on the psychological safety of their teams. Leadership is now distributed in companies and is no longer just a matter of the top.

I have not yet found a company in which there are not, at least in isolated cases, some great, fearless teams with climates for speaking up openly in difficult situations. But in the same company, elsewhere, you find teams with very low psychological safety.

If companies allow more self-organisation and leadership becomes more distributed in the team, does that promote or hinder psychological safety?

It depends. In holacratic, teal or otherwise self-managed, flatter organisations, everyone has to lead themselves and their colleagues. Leadership is not about being someone's boss. Everyone must be self-aware, to know what impact he or she has on others. Not everyone has the skills to do this. But those who do also promote psychological safety. It is not a boss who is in charge, but the rules and structures. In a way, the customer is in charge, indirectly, in that what the work itself needs from people is what drives action.

We need rules and structure

Why are rules and structures helpful?

Everyone agrees that they owe it to each other and to the client to follow certain procedures. These include: if there is a problem to be solved, we have thoughtful conversations about it. There is someone who takes care of the problem and in the end is responsible for testing or executing it.

Is psychological safety also a question of personality or an individual competence?

Psychological safety is not a personality factor; technically it describes an emergent property of a group. The interpersonal climate in a group is co-created by many factors, including the behaviors of leaders and others.

So, can psychological safety be learned?

Yes, anyone can learn how to contribute to a more open, psychologically safe climate in their team. It starts with taking small risks, such as by pointing out the novelty or uncertainty of the work that lies ahead, or by asking colleagues for their ideas, concerns, and questions. Then by listening thoughtfully to what others say, you help build an environment where people begin to realize their voice matters.

Anyone can learn how to contribute to a more open, psychologically safe climate in their team.

In teams, people tend to follow the prevailing opinion. Many people are not aware that they act this way – out of a hidden fear of isolation. What can be done about this?

This tendency is often referred to as groupthink. This is a well-documented phenomenon. It doesn't mean that we all just spontaneously think alike, but rather that our desire to be a part of a group leads us to suppress dissent. This can even lead us to almost feel like we agree, even though in the back of our minds we worry that it might be wrong. Leadership can help overcome this - through a handful of simple interventions.

For example, in meetings you can appoint someone from the group to be the devil's advocate. Whoever is assigned this role should explicitly take a different view from the group. This person thus takes a different view without running the risk of being excluded from the group - after all, he or she has been assigned the task.

Another tool is the so-called pre-mortem analysis. If a group agrees on a risky or uncertain decision, it can pause and say: Let's imagine we’re in the future, and unfortunately our project was a fiasco. Why did this happen? Who can think of the best reason? For example, because “we only asked three clients for input.”

Another approach is to introduce consent instead of consensus for uncertain decisions. This means that everyone doesn’t need to agree with the decision, but they do need to agree that it’s safe to try out a certain thing. This is how you avoid the trap of falling into an obvious failure, while still having a bias toward action in the face of uncertainty. This means a group cannot block an idea without good arguments, and allows people to design useful experiments of uncertain ideas.

Consent instead of consensus

One tip you often give leaders is to ask questions. But when you ask questions, you also have to admit that you don't know everything - in other words, show vulnerability. How does that fit with the traditional image of leadership in organisations?

All leaders are vulnerable. Financial challenges, climate change, a remaining global pandemic, just to name a few, are sources of vulnerability.  There are external events that are bigger than the leaders themselves. The question is whether they are willing to acknowledge it. When they name problems, they give others permission to talk about them too. Vulnerability does not mean false modesty or weakness. On the contrary. Especially when one is ambitious and eager to perform, one dares to ask: What are we still missing? Who has a different perspective? What possibilities do we have for action? These are not suggestive questions that already contain a solution, but what or how questions.

All leaders are vulnerable. Financial challenges, climate change, a remaining global pandemic, just to name a few, are sources of vulnerability.

If the challenges are great, so are the goals. In what way can this affect psychological safety?

It is bad for psychological safety when you have challenging goals and don’t admit it or name it aloud.  But challenging goals that are explicitly named as such are seen by others as an invitation to contribute – to help accomplish them.

Ultimately, companies are about performance. How are psychological safety and performance related?

Data in thousands of peer-reviewed academic studies consistently show a link between psychological safety and performance in teams, and in companies. In an uncertain, complex world where knowledge, work and uncertainty are intertwined, you need psychological safety to perform. Because performance requires learning and that is what it encourages.

Psychological safety and performance are related

Well, in these times we hear a lot about Elon Musk, who, among other things, fired a developer who dared to contradict him publicly. How do you think this company will perform?

Nothing is more foolish than firing someone who corrects a mistake of the boss. When that happens, smart bosses are grateful – and they reward the behaviour to ensure that it keeps happening. And this is especially true if you are not an expert in the field. Your example of developer is a story where the developer had expertise in the coding; he had developed it. Musk did not have that expertise. Of course, it is human to want to be right about everything. But I have seen many companies where this dynamic has led to terrible failures. I'm not optimistic about any company’s future performance if this pattern continues.

So, you think psychological safety is a prerequisite for performance. Can you back that up?

I did a study in the pharmaceutical industry. Every team had an ambitious innovation goal. People wanted to achieve something. To do that, they had to work on innovation projects together with other experts. But the only ones who actually had a breakthrough were those with high psychological safety.

How can you measure psychological safety?

There are different measures that you can integrate into employee surveys. For example, an item that asks: If you make a mistake, will it be used against you in this team? Or: Is it easy to ask for help when you don't know what to do?

As a leader, ask yourself: How much of what I heard this week was good news or bad news, progress or problems, approval or disapproval?

But you can also measure psychological safety informally. As a leader, ask yourself: How much of what I heard this week was good news or bad news, progress or problems, approval or disapproval? If you are too much on the rosy side, chances are pretty good that you don't have enough psychological safety. Because something always goes wrong in an uncertain world; the question is, do you hear about it? If you want to perform, you need to know the truth.

There is nothing harder than psychological safety

The term psychological safety sounds so nice and fluffy. Like, we talk nicely to each other and then everything will be fine. Does that lead to misunderstandings?

Yes, that's my fault (laughs). The term gives the wrong idea. There is nothing harder than psychological safety. Openness is not easy. It’s not soft. You can get it wrong by either making it quite difficult to speak up. Or you can demand openness too much, so that your push into the land of openness becomes rough and inhuman – and therefore dies rather quickly.

There are companies like Netflix that cultivate a culture of absolute openness. At the same time, in a way they are brutal – in that they always discuss even small mistakes publicly or fire people easily. Is openness always good?

I am not a fan of brutal candour. Brutality will never create safety. Psychological safety means permission to be open. But if in our candour we routinely do things that are hurtful, and don’t take responsibility for that, then candour will diminish. So, we will destroy the very thing we are trying to create. Radical openness, as Kim Scott describes it, comes from a position of humility.

It's not fun to hear that we didn't do something well or were ineffective. But we can train ourselves to learn to listen to it.

Unpopular truths are especially difficult to speak when it comes to personal matters. For example, how can you create a safe environment where you can say to people, "We don't need your talents here in the team right now"?

This brings us into the area of feedback. We humans have a real aversion to feedback, except the positive kind. It's not fun to hear that we didn't do something well or were ineffective. But we can train ourselves to learn to listen to it, by developing a growth mindset.

How can we train it?

You can take a lesson from scientists. They don't like it when experiments fail. But they train themselves to cope with failure because they know that there is no way around it if they want to succeed. You have to keep telling yourself, but also others, that you become smarter and more effective when you test your limits – compared to if you always just do what you can already do in your sleep.

Admit failures

Sometimes failure is glorified – at the Fuck-up Nights, for example. But who wants to fail, if we are honest?

No one, that's for sure! Nor do I. I don't even like to trip on the sidewalk. When that happens, I can’t help looking around to see if anyone has noticed. We want to be perfect, but we're not. We are imperfect. At the same time, a certain kind of failure-talk has become faddish. This talk fails to distinguish between truly useful discoveries that some failures bring and simple mistakes that were arguably avoidable. If we celebrate every bad thing caused by sloppiness or perhaps even fraud, that is not healthy. It’s ingenuous, and ultimately it will lead people to hide failures.

How do you fail properly?

First of all, we need to realise that we can fail at what we do. We will occasionally make mistakes. There will be failures that could have been avoided, but we can still learn from them as quickly as possible and move on. Second, recognize that there is a good kind of failure – one that happens due to something that is essentially an experiment. Good failures happen in new territory. We couldn't have looked up the answer because nobody knows it yet. They start with a hypothesis. We test the hypothesis through action, and it turns out that we were wrong. But that's okay, because it's a discovery. We take that discovery and use it for the next step.


About Amy Edmondson

Amy C. Edmondson is a professor of leadership and management at Harvard Business School. Back in the 1990s, she conducted research in hospitals. She found that teams that didn’t experience or expect punishment when they reported errors had a higher error detection rate than others. Her work on this was ground-breaking and shaped the concept of psychological safety. She has also conducted research on teamwork in dynamic work environments. Her books "The Fearless Organisation" and "Teaming" are management bestsellers. Edmondson has been on the twice-yearly global Thinkers50 list of the best management thinkers since 2011.