Česky
Slovensky
EnglishIn the late 1990s, Amy Edmondson studied nursing teams across two hospitals, expecting an obvious result: the teams with the best relationships would make the fewest medication errors. The data showed the opposite. The highest-rated teams reported more errors, not fewer.
For a moment it looked like nonsense. Then Edmondson understood. The best teams were not making more mistakes. They simply were not afraid to talk about them. Where fear ruled, a nurse quietly buried an error and hoped nobody noticed. Where it felt safe, she reported it, the team discussed it, and next time they steered around it.
That inverted relationship named a concept Google would crown two decades later as the strongest predictor of team performance. It is called psychological safety. And if you picture a mellow, conflict-free atmosphere, you have it backwards.
What psychological safety actually is
Edmondson (1999, Administrative Science Quarterly) defined it as a shared belief that the team is a safe place for interpersonal risk-taking. The word "shared" is not there by accident. This is not about how you feel on your own. It is about a norm the group holds together.
Interpersonal risk-taking sounds abstract, but you know it from every meeting. Asking about something everyone else may have grasped ages ago. Admitting you got it wrong. Disagreeing with your boss in front of the room. Saying "I don't understand this" or "I think we're mistaken." Each of those moments carries the risk of looking ignorant, incompetent, or like a troublemaker.
In a safe team you take that risk, because you expect nobody will humiliate or punish you for it. In an unsafe one you stay quiet. And silence at work is expensive, because a problem nobody talks about is a problem nobody solves.
Edmondson named four fears we instinctively dodge on the job. We do not want to look ignorant, so we skip the question. We do not want to look incompetent, so we hide the weakness. We worry about seeming intrusive, so the idea stays in our head. And we do not want to be the eternal critic, so we let a decision that bothers us slide. Every one of those fears makes sense in the moment and robs the team of information over time.
Watch out for one common mix-up. Safety does not mean you are buddies with your colleagues, or that you are gentle and pleasant with each other. A team can genuinely like one another and still dread the hard truth. And a team that argues fiercely over substance can be extraordinarily safe, as long as that argument costs no one their standing or respect.
Google wanted proof: Project Aristotle
In 2012, Google launched an internal study with an ambitious goal: figure out what makes a team great. The project was named Aristotle, and over roughly two years it worked through about 180 teams across the company. The question was simple. Why do some teams excel while others, stocked with equally smart people, fall behind?
The researchers first bet on composition. Pair up clever introverts, or blend introverts with extroverts, or connect people who share interests. None of those hypotheses held up. It turned out that who sits on a team matters far less than how those people treat each other.
Five factors separated the best teams from the rest: psychological safety, dependability, structure and clarity, meaning in the work, and the belief that it has impact. And one of them towered over the others.
Psychological safety came out as by far the most important of the five factors, and at the same time as the foundation the other four rest on.
The logic behind it is plain. Without safety, people will not admit they are falling behind, so dependability collapses. They will not say the brief is unclear, so structure collapses. They stay quiet when the meaning drains out of the work. Safety is not one of five independent factors. It is the soil the others grow in.
Safety is not comfort
This is where a lot of managers go wrong. They hear "psychological safety" and translate it as "let's be nice and never push anyone." The result is a team where nobody is afraid and nobody works hard either. Edmondson has spent years correcting this misunderstanding, and in The Fearless Organization (2018) she uses a simple matrix to do it.
Set two axes against each other: psychological safety and standards, meaning accountability for results. You get four quadrants.
| Low standards | High standards | |
|---|---|---|
| High safety | Comfort zone: people speak openly, but no result is pushing anyone | Learning zone: openness plus a real drive to deliver |
| Low safety | Apathy zone: nobody engages, work gets phoned in | Anxiety zone: pressure to perform without the nerve to tell the truth, a path to burnout |
Aim for the top-right corner, the learning zone. High safety paired with high standards. People chase a demanding goal and still feel free to admit it is not going to plan, ask for help, or question the direction.
The gap between the comfort zone and the learning zone tends to be subtle and brutal at once. In both, people talk openly. Only in the second does a demanding goal sit behind that openness, forcing the team to actually put it to use. Safety without standards manufactures a pleasant meeting culture, all warm sharing and nothing shipped.
That comfort zone is a treacherous trap. It looks healthy, everyone feels fine, yet the team goes nowhere. And the anxiety zone is exactly Edmondson's hospital before safety arrived: high pressure, no willingness to admit an error, so the errors slide under the rug until they blow up.
How a leader builds it
Psychological safety cannot be declared in a meeting. The sentence "from now on it's safe here, feel free to speak" does nothing, because safety is inferred from behavior, not from announcements. People watch what happens to whoever speaks first. Then they arrange themselves accordingly.
Take a typical scene. On the morning standup, a junior developer named Alex says he pushed a change that took down part of the site for twenty minutes. The lead has two options. Either "how on earth did you let that happen" in front of the whole team, or "thanks for flagging it right away, what did we learn and how do we catch it sooner next time?" The first reaction guarantees the next screw-up gets hidden. The second turns a mistake into raw material for improvement. The difference is one sentence, and it shapes the team for months.
One thing about this is merciless. Safety is built slowly and demolished fast. You can spend half a year patiently inviting people to open up, then flatten it for months with a single joke at someone's expense. The team remembers that one reaction, not the twenty welcoming ones before it.
Edmondson and the Aristotle data agree on a handful of behaviors that build safety more than anything else:
- Admit your own fallibility. When a leader says "I read this wrong" or "I'm not sure what to make of it, I need your input," it gives everyone else permission to do the same.
- Ask more than you answer. The question "what might I be missing here?" at the end of a meeting opens a door that ten calls for openness never will.
- Your response to bad news decides everything. Someone brings you a mess, you explode, and you have just shut the door for next time.
- Keep an honest experiment separate from carelessness. A failed attempt the team learned from is not the same as negligence, and punishing both the same way teaches people to try nothing new.
- Thank people for the courage, not just for good news. "Thanks for saying it out loud" honors the act of speaking, even when the content stings.
Responding productively to bad news is really feedback in real time. Safety and the ability to give and take feedback feed each other, and we broke the mechanics down in our guide on how to give and receive feedback.
How to tell where your team stands
Safety can be measured. In her 1999 research, Edmondson used a seven-item questionnaire that people answer on an agreement scale. You do not need to run it formally. Just walk through the items in your head for your own team and be honest while you do. A few, paraphrased:
- When you make a mistake on this team, does it get held against you? The more often yes, the worse.
- Can you raise a hard topic or a problem here without getting burned for it?
- Do you feel safe sticking your neck out and taking a risk?
- Is it easy to ask colleagues for help, or does asking feel embarrassing?
- Does the team value your specific skills, or overlook them?
When was the last time you said "I don't understand this" in a meeting? And when did you swallow it instead, because keeping quiet felt safer? Your answer reveals more about the team than any questionnaire.
You can spot the warning signs without paper, too. The same two people talk in every meeting. Nobody ever asks a dumb question. Mistakes only surface once it is too late to fix them. And when the boss asks "does anyone have concerns?", silence follows, after which the same people say something entirely different out in the hallway.
What a regular team member can do
Most writing on psychological safety aims at managers, as if a rank-and-file employee were just a passive recipient of the atmosphere. That is not how it works. The norm forms from the bottom as much as from the top, and one brave person can raise the bar for the whole team.
When you speak up with a question or own a mistake, you do two things at once. You solve your own problem, and you show everyone else the world did not end. Next time someone else speaks up. Safety is contagious both ways, because fear spreads and so does courage.
Protecting others who take a risk is within your reach as well. When a colleague floats an unpopular view and the room piles on, it takes one sentence: "hold on, that idea is worth a look." When someone asks about something basic, do not signal that it is beneath you. These small reactions decide whether people speak up next time or crawl back into their shells.
It helps to know your own tendencies. On a team, one person naturally takes on the role of asking the awkward questions, while another instinctively softens conflict and guards the mood. Neither is wrong, but each shapes the safety around it differently. If you want to see which roles you lean toward, a team roles test will give you a read.
Edmondson leaves us with one uncomfortable note: on a team without safety, people are behaving completely rationally when they stay quiet. Silence costs nothing, speaking up can cost a lot. Until someone flips that equation, whether the leader through her reaction to bad news or a colleague by speaking first, no meeting about company values will move it an inch.