The leadership team meets monthly. A product manager raises a risk that could derail the roadmap, the chief executive thanks her warmly for her candour, and the topic quietly disappears from every future agenda. Nobody was punished, and nobody volunteered bad news again for a quarter. Psychological safety was absent precisely because the first response taught the room what honesty costs.
Amy Edmondson's research frames psychological safety as a shared belief that the team is safe for interpersonal risk taking: people can ask naive questions, admit mistakes and challenge plans without fearing embarrassment or retaliation. It is one of the most consistent team-level predictors of learning behaviour we have, and sponsors increasingly ask coaches to produce it, ideally with a number attached.
Here is the uncomfortable part: you cannot coach a property of a system the way you coach an individual skill. You can, however, work the levers that move it. This article covers the leader behaviours that matter, the team experiments that make speaking up routine, the instrumentation that shows movement, and the way your own conduct in the room teaches more than any model.
Why psychological safety resists direct coaching
Psychological safety is not a trait residing in a person. It emerges from history, interdependence and above all from how leaders respond when someone takes a risk. Two teams in the same company, doing the same work under different managers, can have entirely different climates. That is why a team-level property cannot be developed the way you would develop an individual's feedback skills.
This creates a trap for coaches. If a sponsor says "fix my team's silence" and you respond by coaching the quietest member to be braver, you have treated a symptom and quietly blamed the patient. The system will re-create the silence, and the person you coached now carries extra risk with no extra cover.
The practical consequence is contractual. Coach the leader whose behaviour prices honesty, or work with the team's routines where speaking up either happens or does not. Define success as observable behaviour change, such as dissent appearing in meetings and decisions being reviewed afterwards, not as a feeling somebody reports having. Vague goals produce vague climates; specific behaviours can be practised, observed and reported.
Key takeaway: you cannot coach psychological safety directly; you coach the behaviours and routines that make honesty cheap.
The leader behaviours that move the number
Explicit invitations come first. Leaders believe an open-door policy is an invitation; teams know it is furniture. What actually produces speech is a named, specific invitation at the moment it matters: "Before we agree, I want the strongest case that this plan fails. Who sees it?" Silence after a general plea for honesty reads as danger; silence after a direct question reads as an answer to that question.
The second behaviour is the response to bad news, which sets the exchange rate for honesty across the whole team. Appreciation comes before evaluation, curiosity before verdict. The messenger should leave the conversation more willing to return next time, because everyone else in the room is silently checking whether they would come back.
Fallibility modelling is third. A leader who says "I might be wrong about this" gives everyone else permission to be uncertain too, and most leaders seriously underestimate how much their certainty suppresses the room. Framing work as a learning problem rather than a pure execution problem completes the set, which is why these behaviours sit naturally inside the four coaching conversations managers already have. A leader who practises those conversations deliberately is, in effect, running a psychological safety programme without the banner.
Signals and instrumentation: making the change visible
Because psychological safety is qualitative, instrumentation means finding honest proxies. Short team surveys built on validated team-learning items are the standard starting point: a handful of statements about asking questions, raising problems and challenging ideas, scored as a pulse rather than a verdict. Trend matters more than level, and anonymity protects the signal from politeness.
Meeting behaviour counts are the underrated instrument. Tally, across a sample of meetings, how many questions are asked, how long the leader speaks, who talks after the leader states a view and whether any dissent survives to the decision. These counts are crude, observable and remarkably hard to game, which is exactly what makes them useful alongside a survey.
Decision reviews add a third source: after significant calls, record whether contrary evidence was voiced, who voiced it and how it was treated. For leaders, a structured 360 view is often the mirror that makes the pattern undeniable, and a 360-degree feedback rollout for leaders gives you the mechanism. Teams in volatile conditions need this most, since turbulence raises both the value and the cost of candour, a point developed further in leading teams through uncertainty without panic.
Team-level experiments a coach can run
Experiments beat exhortations. A speaking-up retrospective asks one question at the end of a real meeting: what did we almost say this month and decided not to? The first round will produce silence, which is itself the data. Run it monthly, thank the first speaker extravagantly, and watch the answers lengthen as the room learns the question is genuine.
A decision review ritual goes deeper. Once a month, the team picks one significant decision and answers three questions: what did we believe, what did we choose not to look at, and whose doubt went unspoken? Handled without blame, the ritual normalises dissent as maintenance rather than disloyalty, and it produces a written record a coach can work with.
Norm experiments change the meeting mechanics themselves. The leader speaks last on contested items, a round-robin precedes open discussion, and a rotating red-team challenges the plan before commitment. Each experiment is small, time-boxed and reversible, which is precisely what makes it safe to try. For the routines that hold between meetings, the team-building and high-performance course extends this work into norms and operating rhythm. Measure with your meeting counts and pulse surveys, and let the team see their own numbers, because data produced together belongs to them.
The behaviour-to-experiment map
The table below connects the leader behaviours that drive psychological safety to the team signals worth watching and the experiments that respond to them. Use it to turn a vague request for "more psychological safety" into a contract with visible moves.
| Leader behaviour | Team signal to watch | Coaching experiment |
|---|---|---|
| Explicit invitations | Silence after open questions | A named, specific invitation on the riskiest agenda item, every meeting |
| Response to bad news | Bad news arrives late or never | Retrace the last suppressed risk and redesign the leader's first response |
| Fallibility modelling | Leader talks most and hedges least | Leader opens one meeting a month with where they might be wrong |
| Learning framing | The same failure repeats without discussion | Monthly decision review recording dissent and how it was treated |
| Equal airtime | Two voices dominate every discussion | Round-robin before open debate on contested items |
Pick one row, not five. Behavioural change in leaders is expensive, and a single row practised consistently beats a poster of all five attempted once.
Model what you coach
Your own conduct is part of the intervention. How you respond when a client pushes back, admits a failure or disagrees with your framing teaches the room what candour buys in your presence. Coaches who demand speaking up from teams while bristling at challenge in sessions get found out quickly, and the finding spreads.
Triadic work adds confidentiality edges. Agree themes-not-content reporting with both parties before the engagement starts, and say in advance what you will do when sponsor curiosity and client privacy collide. Those edge-setting conversations are skills in their own right; conflict resolution skills for workplace coaches sharpens exactly that muscle. Triadic contracting as a design discipline is taught in the Leadership Coaching Mastery certification, and it spares you from improvising these rules in the fourth month of an engagement. In practice, the coach who models fallibility first gives everyone else permission to follow, and the team watches how you take your own medicine.
Frequently asked questions
Can a coach really measure psychological safety?
Measure it indirectly and directionally, yes. Short validated team surveys, meeting behaviour counts and decision reviews together give a defensible read on whether speaking up is getting safer. Treat every number as a trend rather than a verdict, pair it with observation and the team's own account, and never present a single pulse score as proof that a culture has changed.
How long does it take to shift psychological safety?
Expect months, not weeks, because the history that created the silence must be overwritten by repeated contrary experience. Meeting behaviour usually shifts first, within a few cycles of consistent leader behaviour, while survey trends lag behind. If nothing moves after a quarter of genuine leader effort, look for an unaddressed episode that recently taught people the opposite lesson.
Can individual coaching fix a team's psychological safety?
Not on its own, because psychological safety is a property of the team system rather than a quality of one member. Individual coaching helps most when it targets the leader, whose responses set the price of honesty. Pair any individual work with team-level routines such as decision reviews, or the system will quietly re-create the silence you were hired to remove.
What if the sponsor wants a score but no behaviour change?
Push back before you contract. A number without behaviour change invites gaming and teaches the team that measurement is theatre. Re-contract around two or three observable leader behaviours and the team routines that accompany them, report trends in those, and let any survey numbers follow the behaviour rather than substitute for it. Scores chase behaviour; they never lead it.
Your next step
Run one speaking-up retrospective with a team you coach this month, and record meeting behaviour counts before and after. Bring the behaviour-to-experiment table to the sponsor conversation and agree which row you are coaching first. For the leader-side skills, the Leadership Coaching Mastery certification covers triadic contracting and 360 debriefs, and the team-building and high-performance course extends the team-routine work between sessions.