Psychological Safety Is Not Comfort: Why It Makes High Standards Possible

If psychological safety is something you're paying attention to, you're paying attention to the right thing.

I'd go further. Psychological safety sits close to a foundational need for any team. People arrive at work wanting to know whether it's safe to speak, safe to be wrong, safe to be themselves in front of the group. When the answer is yes, most other problems become workable. When the answer is no, something else moves into the space, and it is never neutral. Fear, most often. Sometimes apathy, because caring has become too expensive. Sometimes anger, from the people who have decided they have nothing left to lose.

So the definition matters.

Somewhere between Amy Edmondson's research and the average leadership offsite, the term arrived as a mood. Be kind. Be inclusive. Make sure nobody feels uncomfortable.

The result is a false choice. Pick safety and you stop saying anything hard. Pick standards and you treat safety as an obstacle to performance. Both get the relationship backwards. Low safety does not produce high standards. It produces buried problems and quiet turnover. A standard nobody dares question is not a standard. It is compliance, and compliance stops working the moment nobody is watching.

Psychological safety is what makes a standard enforceable. It exists so hard things can be said in both directions, rather than avoided in both directions.

One caveat before we start. This is my own read rather than research. There are people who understand the mechanics far better than I do, and Edmondson's writing is where to go for the evidence base. What follows is what I have seen work, and what I have seen go wrong.

What is psychological safety, really?

Psychological safety is the shared belief you can take an interpersonal risk on this team without being punished or humiliated for it. Raising a problem. Admitting an error. Disagreeing with your manager. Asking a question when you do not know the answer.

The concept comes from Amy Edmondson at Harvard Business School, whose early work looked at medication errors in hospital teams. She expected the better-performing units to record fewer errors. The data went the other way. Stronger units reported more.

They were not making more mistakes. They were willing to say so. The weaker units were making the same errors and absorbing them quietly.

The finding is the whole concept in one line. Safety changes what gets reported, not what happens. Google's Project Aristotle later put the term in front of business audiences, identifying psychological safety as the strongest of the team dynamics it measured.

The clearest way to hold the idea is against the things it gets mistaken for.

• Candour about risks, doubts and mistakes. Not politeness, which leaves the real issue unsaid.
• High standards, clearly stated and consistently held. Not a free pass on weak work.
• Disagreement in the open, aimed at a better decision. Not an absence of disagreement, bought by avoiding the conversation.
• What the team as a whole expects to happen when someone speaks up. Not one person's trust in their manager, or how kind the manager seems.

What are the four zones of psychological safety?

Safety and standards are two separate dials, and where your team sits depends on both. Edmondson maps the four combinations in The Fearless Organization, setting psychological safety against accountability for high standards. She calls them zones, and only one of them performs.

The grid is easier to recognise than to describe. Your zone shows up in small things rather than in anything you have declared. Who speaks first in your meetings. What happens when a piece of work is late. Whether a risk gets raised or quietly logged. What follows is what each one looks like from the inside.

Anxiety. Low safety, high standards. The project status stays green until the fortnight before delivery, then turns red overnight. Three people knew in month two. The reporting was accurate about the plan and silent about reality. A high standard here stops being a shared ambition and becomes a threat, so people stop volunteering for hard work and stop showing you early drafts. Teams land here under competent, well-intentioned leaders who have never been told which of their habits close a room.

Comfort. High safety, low standards. The same three action items roll over four meetings running and nobody mentions it. One person's work has been below standard for a year, and your team has built a workaround rather than had a conversation. You leave the meeting feeling it went well, and the same list comes back next fortnight.

Apathy. Low safety, low standards. You ask what someone thinks and get "whatever you think". Nobody applies for the internal roles. Handovers happen by accident, and nobody chases the gap.

Learning. High safety, high standards. Someone junior challenges a senior estimate in your meeting and the estimate changes. Bad news reaches you early, while it is still small. The argument is about the work rather than about who will carry the blame for it.

Anxiety and comfort share a mechanism. Your team manages you, and you read what comes back as the truth.

Two things are worth drawing out of the grid. Apathy is rarely where a team starts. In my experience it is where anxiety and comfort both end up when nobody intervenes. Each loses the one thing it had going for it. Anxiety loses the standard, because people work out caring is too expensive and drop back to whatever avoids trouble. Comfort loses the safety, because safety never tested on anything difficult turns out to have been politeness. Apathy is also far harder to reverse than the failures leading to it, because the people who would have pushed back have already stopped bothering.

I have been on the wrong side of this, not as an observer but as the person who left.

The pattern was consistent. Ask a question in private and it registered as a challenge. Ask it in a meeting and whoever asked was taken down in front of the room, so the rest of us learned quickly which subjects carried a cost.

What followed is what the grid predicts. The people who stayed narrowed to the minimum required. I went from a strong performer on the team to resigning. It was never named as a safety problem. It was recorded as a resignation.

It did not stay at work either. It had me questioning my own ability and my value. I still catch myself doing it.

Working on safety or standards alone does not get you there. Set out to make people feel safe, drop the standard to do it, and you land in comfort. Set out to lift performance by adding pressure, skip the safety, and you land in anxiety. You have moved, but not to the zone you wanted.

How do you tell where your own team sits?

Look at what people do when they have something bad to say, not at what they tell you about the culture.

Seven questions are usually enough.

• When did someone last tell you your plan was wrong, in a meeting, with other people present? If you can't name the occasion, it hasn't happened.
• Are you hearing about problems earlier than you were six months ago, or later?
• Which topics does your team route around? If you can't name one, ask someone who would know.
• How many of the risks raised in your last project review had been known to someone on the team for more than a month?
• What happened to the last person who raised an unpopular concern, and where are they now?
• Do your quietest people speak in meetings, or only in one-on-ones?
• When you ask for input, does anyone go first without being named?

The question about quiet people needs care. Silence is not proof of a problem. People stay quiet because they think before speaking, because the meeting is too big, or because an earlier manager taught them the cost and they have carried it into this job. None of it is your doing.

It is still your problem. A room where three people talk gives you the judgement of three people, and your data set has a hole in it either way. The fix for someone who is naturally quiet differs from the fix for someone who is afraid, and you find out which by asking rather than by watching.

What do these failures cost the business?

Anxiety and comfort both cost you decision quality and both cost you people. They fail in different directions.

• Decision quality. Comfort loses the objection nobody voiced. Anxiety loses the objection nobody dared voice. You paid for the judgement of eleven people and received the judgement of the three who find talking easy.
• Speed. A risk raised in week two costs you a conversation. The same risk raised in week ten costs you a replan, a client call and somebody's weekend. Comfort defers it because raising it feels rude. Anxiety defers it because raising it feels dangerous.
• Retention. Unsafe teams lose people first, and lose the ones with options. Comfortable teams lose them slower, and lose the person who has quietly absorbed three other people's work and worked out the standard was decorative. Neither tells you the real reason in the exit interview.
• Exposure. The failures reaching your board are seldom failures of knowledge. Somebody knew, and you had built no cheap way for them to say so.

An engagement score will pick up some of this. What it will not do is tell you which of the two failures you are looking at, or what to do about it on Monday. The rest arrives in the operating result a year or two later, by which point it looks like a run of unrelated incidents rather than one condition.

How does psychological safety get misused?

In two directions, and both start from the same mistake, treating safety and standards as a choice rather than a pair.

By the leader. Someone takes "safety is not comfort" as confirmation of a virtue they already prized, and starts calling bluntness candour. Directness is about the content of what you say. Bluntness is about what it costs the person to hear it. The first is a service. The second is a fee you charge someone for information they needed anyway.

Five tests separate them.

• Did the person leave with something specific to act on, or only with a clearer sense of your disappointment?
• Would you have said it the same way, in the same tone, to your own manager?
• Are you being precise about the work, or characterising the person?
• Did you ask anything, or only tell?
• Was the timing chosen for their benefit, or for your relief?

Be hard on people and call it high standards, and you produce the outcome described earlier. Filtered information, defensive reporting and quiet departures. Nothing in this argument defends it. It is the point made in Empathy With Boundaries, where care and clarity are the same act rather than competing instincts.

By the team member. The term also gets used as a reason not to hear something. A performance conversation gets reframed as an attack. Disagreement gets read as disrespect. The vocabulary of safety attaches itself to ordinary discomfort, and a manager who has done nothing wrong starts avoiding the conversation anyway.

This one is harder to raise, because the genuine version is far more common than the misuse and nobody wants to be the person who waved away a real complaint. So test the situation rather than the word. Was the person given accurate information about their work? Were they given it early enough to act on? Was it about the work rather than about them? Three yeses and the discomfort is the job rather than a breach.

The cost falls on the people who need the term most. Once "I do not feel safe" becomes unanswerable, managers stop telling the difference between someone being harmed and someone being challenged, and the concept protects nobody.

What builds psychological safety?

Behaviour in the moments where speaking up is costly. Nothing else moves it. Not a workshop, not a values poster, not a survey.

This is the part which gets left off the slide. Safety is expensive for the person with the most power in the room. Everyone else gets it cheaply, and only after you have paid.

Seven moves do most of the work.

Go first, on the record. Name your own error before anyone else has to. Make it specific and make it recent. Vague self-deprecation reads as performance. The same goes for the questions you do not know the answer to. Ask them in the room where you are meant to be the most senior voice, and the team learns it is allowed.

Separate the report from the fault. When someone brings you a problem they created, respond to the reporting before you respond to the mistake. Acknowledge the speed, then deal with the substance. Reverse the order and you have trained them to wait until it's unavoidable. Every reaction you have is a data point, and one sigh at the wrong moment buys you six months of late reporting.

Ask for the disagreement by name. "Any concerns" gets nothing. "Priya, you've run this integration before, where does this plan break" gets something real. Broad invitations are answered by the confident. Direct ones are answered by the informed.

Accept being visibly persuaded. Changing your position in front of your team because someone made a better argument is the single strongest signal available to you. The instinct is to do it once, privately, then announce the revised decision as though you arrived at it alone. The lesson lands backwards.

Close the loop. Tell people what happened with the thing they raised, including when the answer is no. Raising something and hearing nothing back teaches a team the exercise is decorative.

Deal with the room, not only yourself. Your own behaviour sets the ceiling. Other people set the floor. You can run a faultless meeting and still have an unsafe team, because the cost of speaking is being charged by a peer rather than by you. The eye-roll when a particular person talks. The expert who makes a question feel stupid. The colleague who repeats a private admission in public. Everyone in the room can see it. Everyone is waiting to see whether you will. Name it once, in the moment, and the room learns the standard covers conduct as well as work.

Hold the standard while protecting the person. Be precise about the work and generous about the human. Vague criticism of a person feels like an attack. Specific criticism of work feels like respect, and it gives them something to act on.

The last one is a coaching skill rather than a personality trait, so it can be taught. The mechanics are covered in more detail in Coaching Your Team.

None of it survives an incentive pointing the other way. Where managers get promoted for having no visible problems, you will produce leaders who use the vocabulary fluently and suppress the same information. The pattern is described in Why Most Leadership Development Fails. First-time managers pick up the local version fastest, since they learn by copying whoever is nearest, as set out in How to Set Up First-Time Managers for Success.

Does holding someone to account break psychological safety?

No. Being held to account is not a breach of safety. Being ambushed is.

A team where underperformance is never addressed is not a safe team. It's a team where you traded the standard for a quiet life, and your strong performers know it.

The safety test in a difficult performance conversation is not whether the person enjoys it. It's whether they saw it coming, whether they were told early enough to do something about it, and whether they leave the room knowing precisely what to change. Handled well, accountability builds safety rather than spending it.

The distinction sits in the preparation rather than the tone. A conversation someone has been set up to succeed in feels different from one they've been set up to lose, and everybody in the team can tell which kind you run. More on structuring them in How to Run a Performance Review.

Frequently asked questions

Is psychological safety an excuse for poor performance?

No. Safety is permission to take an interpersonal risk, not permission to miss the standard. The two get conflated because both involve a leader tolerating something uncomfortable. Tolerating a hard question is safety. Tolerating sustained underperformance without naming it is avoidance, and the team reads it as the standard being optional.

Is psychological safety the same as trust?

No. Trust is about whether you rely on a specific person. Psychological safety is about what the whole team expects will happen when someone speaks up. Someone can trust you completely in a one-on-one and still decline to raise a risk in a team meeting, because the room is not you. There are eleven other people in it. Safety gets built in public rather than through private reassurance.

Can you have psychological safety in a high-pressure environment?

Yes, and high-pressure environments need it more. Edmondson's original work was in hospitals. Safety is not the absence of pressure. It's the absence of interpersonal risk when reporting under pressure, and it is precisely what stops small failures compounding.

How do you measure psychological safety?

Survey instruments give you a directional read, and Edmondson's original seven-item scale is still widely used. Treat behavioural indicators as the better evidence. Track how early risks surface, how often decisions are challenged in the room, and how many people contributed in the last four meetings.

The reframe worth keeping

Teams rarely fail because nobody knew. They fail because nobody judged it worth saying.

Psychological safety does not lower the standard. It is what lets the standard survive contact with reality.

The combination is the job. Be precise about the work, generous about the person, and consistent enough for neither to come as a surprise.

The hard part was never the language. It is holding both at once, in the moment, when the conversation is uncomfortable and you would rather have it next week. At Catalyst we build this into manager practice through the CORE framework. Context, Observation, Response, Evolve. It gives your managers a repeatable structure for the conversations most of them avoid.

If this is the gap in your organisation, book a discovery call and we'll work out what's missing.