Why your psychological safety initiative is not working.

You ran the session. People nodded. Someone wrote "speak-up culture" on a whiteboard. And six weeks later the meetings sound exactly the way they always did.

This is the most common version of the problem I see. A leader invests in psychological safety, watches nothing change, and quietly concludes the whole idea is soft. The idea is not soft. The mistake is treating safety as a general mood you can lift across the board, when it actually works as a set of specific permissions. Some of those permissions are easy to grant. One of them is hard. And the hard one is almost always the one your initiative never touched.

Here is what that looks like in real data.

Safety is not one thing. It is five.

We recently ran a psychological safety assessment across an organization. 171 people answered. The instrument measures five separate dimensions, not one overall feeling, and that detail turns out to matter more than anything else in this piece.

Four of the five held up. People said problems get discussed directly, and that one was the strongest of all: roughly three in four respondents affirmed it. People said they feel heard in conversations. People said they feel supported by their colleagues. The dimension on clear processes and transparent decisions came back more mixed, but still on the positive side of the line. By any reasonable read, this is a capable, well-meaning organization full of people who like and respect each other.

Then there was the fifth dimension. The one about trying new things, taking risks, and admitting mistakes.

On that question, only about a third of people (36%) said it happens often or always. More than a quarter said it happens rarely or never. It was the only dimension on the entire assessment where the positive responses were a minority. Same people. Same teams. Same week. A gap of more than forty points between how safe it felt to discuss a problem and how safe it felt to take a risk or own a mistake.

That gap is the whole story. Safety had not failed everywhere. It had failed in exactly one place: the moment something might go wrong.

What the low score was actually telling us

A score like that is not random, and it is not a character flaw in the people. It usually has a mechanism. In this case, the mechanism showed up in a second set of data.

We had also run a personality assessment across the leadership group. The profile skewed hard toward fast-moving, results-driven styles. A lot of drivers. People who are genuinely good at deciding quickly and pushing toward deployment. That is a strength. It is probably part of why the organization succeeds.

But put a room full of drivers in charge of a risk conversation and a pattern emerges. An advisor raises a flag. This will take longer than planned. This needs more resources. This carries a risk we have not sized yet. And the instinct of a driver, under time pressure, is to keep the train moving. "Don't worry about that. We'll handle it when we get there."

Then the risk arrives, the way unsized risks do. The timeline slips. And now senior leadership is frustrated, because from where they sit the work looks slow. So the pressure to move faster goes up, not down. Which makes the next advisor even less likely to be the person who stands up and says wait.

Do that once and nothing happens. Do it across a year and the advisors learn the lesson. Flagging a risk costs you something and changes nothing. So they stop. Not loudly. They just go a little quieter each time. The risk still exists. It simply stops being said out loud, which is the worst possible place for a risk to live.

That is what a low score on the risk-and-mistakes dimension actually measures. Not weakness in the people. A rational response to what keeps happening when they speak up.

The two words that close a room

When we watched these leaders in simulation, playing out the exact conversations they have at work, the pattern had a sound. Two words.

"Yeah, but."

An advisor would raise a concern, and the leader would say "yeah, but" and then explain why the concern did not need to slow anything down. Reasonable tone. No hostility. They thought they were being decisive. What the other person heard was: your input has been received and overridden.

"Yeah, but" is one of the most efficient ways to teach a team to stop talking. It validates nothing and closes the door in the same breath. And the leaders doing it had no idea they were doing it. That is the part worth sitting with. The behaviour suppressing candor was invisible to the people performing it.

What we teach leaders to do instead

The fix is not a poster about courage. It is a different move in one specific moment: the moment someone raises something inconvenient.

Replace the reflex to respond with the reflex to ask. Before you explain why the risk is manageable, get genuinely curious about the risk.

"Walk me through what you're seeing."

"What would have to be true for this to go wrong?"

"What does this need that it isn't getting right now?"

These are not soft questions. They are how a serious leader pressure-tests a risk instead of waving it off. They also do something "yeah, but" never could. They tell the person who spoke up that speaking up was worth it. That is the real currency of psychological safety. Not comfort. Evidence that candor gets taken seriously.

The harder skill is staying in the question when you disagree. When an advisor flags a timeline you think is too cautious, the curious move and the dismissive move feel almost identical from the inside. Both can be quick. Only one keeps the room open. Learning to tell them apart, in real time, under pressure, is the actual work.

Why this is a practice problem, not a knowledge problem

Every leader in that organization already knew they should listen. Knowing it did nothing. The "yeah, but" still came out, because it is faster than a question and it fires before the thinking brain catches up.

You do not fix a reflex with information. You fix it with repetition. These leaders changed when they got to run the conversation, hear themselves say "yeah, but," feel the room close, and then try it again differently with someone who could push back the way a real advisor does. That is the difference between understanding psychological safety and being able to produce it on a Tuesday afternoon when you are behind schedule and someone hands you a problem you did not want.

This is why a workshop on safety rarely moves the number. The number lives in a behaviour, and behaviour is built where it gets used.

If your initiative stalled

If you have already invested in psychological safety and the meetings still sound the same, the concept is not the problem. Look for the one dimension that did not move. It is usually the one where speaking up carries the most risk and your leaders respond the fastest. That is where the work belongs.

That is the kind of conversation we help leadership teams rebuild, and rehearse, before the stakes are real.