WorkshopSkill activity folders

guide

How to build psychological safety on a team

Psychological safety is not warmth — it is whether people will tell you the bad news while it is still useful. It is measurable in two minutes and raisable only by a specific behaviour, not by a slogan. Here is the sequence that works.

What it is, and what it is not

The phrase comes from Amy Edmondson's work on learning in teams (Edmondson, 1999): a shared belief that the team is safe for interpersonal risk. It is often mistaken for "being nice" or "low standards." The opposite is truer — the highest-performing teams are not polite, they are honest. A team where everyone agrees in the meeting and then quietly disagrees in the group chat has a safety problem, not a harmony win. Safety means the mistake gets said out loud, early, when it is still cheap to fix.

Measure before you fix

You cannot raise a number you have not looked at. The Safety Check asks four private questions — can I ask a dumb question? admit a mistake? disagree? take a risk? — and then reveals the team's four bars. The move is not to average them into a score; it is to circle the lowest one. Four people will rarely rate all four the same, and the lowest bar is the single thing worth working on this week. That keeps the conversation honest and bounded: one behaviour, not a culture overhaul.

The behaviour has to come from the front of the room

Safety is granted, not announced. A poster that says "this is a safe space" changes nothing; the leader admitting one real mistake and asking one question they worry is dumb changes everything, because it puts a price on speaking up and then visibly refuses to collect it. That is why the trust folders run the risk first and small: Minefield builds it one direction at a time, Step In If… lets someone reveal something true with no names attached, The Trust Timeline lays out where trust broke and got repaired. The facilitator going first is the load-bearing step in all of them.

One commit, one date, one re-check

End with a concrete behaviour, an owner and a date — not "communicate more openly." Then, in four weeks, run the same four questions again. The delta is the whole point. Psychological safety is not a wall you build once; it is a garden, and unmeasured it quietly goes back to weeds. A re-check date is what turns today's good meeting into a habit next quarter.

The five steps

  1. Measure it before you try to fix it Rate four questions privately: ask a dumb question, admit a mistake, disagree, take a risk.
  2. Circle the lowest bar Four people rarely rate all four the same — the lowest bar is this week's one thing.
  3. Model the behaviour yourself, first The leader admits one mistake and asks one question they fear is dumb.
  4. Commit to one concrete behaviour Not "be more open" — a specific thing, with an owner and a date.
  5. Re-check on a set date Run the same four questions in four weeks; the delta is the point.

Questions people ask

What is psychological safety, in plain terms?

It is a shared belief that you can speak up — ask a question, admit an error, disagree, try something that might fail — without being punished or humiliated for it. The term is Amy Edmondson's (1999): it is not about being friendly or lowering standards, it is about whether people feel safe enough to surface the truth that lets a team learn. A polite team where nobody flags a problem is a low-safety team.

Can you really measure it?

You can get a two-minute read that is useful without being scientific. The Safety Check folder asks four plain questions and shows the team its own four bars and which one is lowest. That is a field measure, not a psychometric instrument — good enough to start a real conversation and to re-check a shift later, not good enough to rank or hire on. Treat it as a prompt, not a verdict.

Which activities actually raise it?

The ones where a small, reversible risk is taken out loud and rewarded, not punished. Minefield builds trust one direction at a time. Step In If… lets people reveal something true with no names attached. The Trust Timeline puts the moments trust broke and got repaired on the table. None of them work unless the facilitator goes first and is visibly willing to be imperfect.

Go deeper