Every leadership team we speak to says it wants honesty. Very few have built an environment where honesty is the rational choice.
That gap is not hypocrisy. It is arithmetic. In any organization, a person deciding whether to raise a problem runs a quick and mostly unconscious calculation: what does it cost me to say this, and what does it cost me not to?
When silence is cheaper, people go quiet. Not the anxious ones. Everyone.
What silence actually costs you
The visible cost is the idea that never gets said. That is the smallest part.
The larger cost is delay. In teams without psychological safety, bad news travels slowly and upward-filtered. The project that is failing gets reported as "on track with some challenges" for two quarters. By the time leadership sees the real picture, the cheap fixes have expired.
The largest cost is the one nobody attributes correctly: your best people leave, and the exit interview says "growth opportunity". What they mean, often, is that they spent two years watching things go wrong and not being able to say so without a fight.
Google's Project Aristotle looked at 180 internal teams and found psychological safety at the top of the five dynamics that distinguished the most effective ones. Amy Edmondson's original work at Harvard showed the mechanism: safety enables the learning behaviour that sustains performance. It is not a soft outcome. It is the condition under which the rest of your investment works.
The distance version
Remote work does not create this problem. It sharpens it in a specific way: everything is written, and writing is permanent.
In a room, a half-formed objection costs almost nothing. You say it, someone pushes back, the moment passes, nobody remembers by Friday. In a shared channel, the same half-formed objection is a searchable artifact with your name on it, visible to people you have never met, retrievable in two years.
So distributed teams develop a particular kind of caution. Messages get drafted and redrafted. Doubts get softened into questions. Disagreements move to private DMs, where they cannot inform the group. What is left in the public channel is agreement, and agreement in a channel is very easy to mistake for alignment.
The other half of the distance problem is asymmetry. Trust at a distance is not evenly distributed: people trust the two colleagues they happen to speak to daily and hold everyone else at arm's length. A silence from a familiar colleague reads as "busy". The same silence from someone in another hub reads as "difficult". Nobody decided this. It is what happens when the relationship has no body behind it.
Four things that are usually mistaken for safety
Everyone speaks in meetings. People can speak fluently for an hour without saying anything that carries risk. Volume is not exposure.
Nobody complains. In a low-safety environment this is exactly what you would expect to see.
We have an anonymous channel. Useful, but it is a symptom management tool. If the only route for truth is anonymity, the named routes are closed.
The leadership team is very open with each other. Common, and misleading. Safety is not a property of the organization; it exists between specific people, in specific directions. It is usually strong horizontally at the top and weak vertically.
Why you cannot install it with a policy
Safety is built from evidence, not from announcement. People update on what happens after someone takes a risk, not on what was said in the town hall.
Which means the only thing that moves it is a lived sequence: someone says something difficult, and nothing bad happens to them, repeatedly, visibly, including when the person who takes it badly is senior.
Manufacturing that sequence inside normal operations is slow, because normal operations are exactly where the stakes make people careful. It happens faster in a context deliberately built to remove the usual signals of rank.
What that looks like in practice
At Lunita Raiz we measure trust and safety as a distinct dimension of the diagnosis, and in the distributed version we measure it twice, the leadership's reading and the team's, because the gap between the two is often the most useful number in the report.
When it is critical, the programme is designed around it. Some of what we use is unusual for a corporate context, and it works for a structural reason: in a ceremony there are no titles. The CEO and the newest hire are in the same darkness, in the same heat, with the same discomfort. Hierarchy does not dissolve because someone declared a flat culture. It dissolves because for two hours it was physically irrelevant.
What comes back with the team is not a feeling. It is a precedent: a shared memory of having said the true thing and survived it.
LUNITA