Sertaç Yıldırım field notes

Home → Leadership

Psychological Safety: Measurable Behaviour, Not Good Intentions

"Everyone can speak freely here." The manager who says this usually means it. The trouble is that the person saying it is not the person who decides whether it is true. That decision is made by the most junior person in the room, and they make it by watching what actually happens.

Sertaç Yıldırım · · 5 min read

In short
  • A safe team is not a quiet team. You will see more disagreement and more reported failures.
  • It is measurable: who speaks first, how late bad news arrives, how often anyone says "I don't know".
  • The first question after an incident decides everything. "Who did it?" versus "how did the system allow it?"
  • The manager admits a mistake first. Not free, but the cheapest investment available.

From the field: trust lost in two minutes

A Friday evening config change took the payment service down for twenty minutes. On Monday morning ten of us joined the incident review. The opening sentence was: "Who pushed this change?"

There was no malice in it; it was an attempt to gather context quickly. But everyone in the room heard the same thing. The person who made the change answered, went red, and said nothing for the next forty minutes. The action item that came out was "we will be more careful with changes" — in other words, nothing.

The real information arrived three weeks later, in a 1:1: the same config file had caused a similar problem two months earlier and nobody had reported it. So the incident was not a one-off, it was a repeat. We had not learned, because the first one was never discussed.

Blame does not reduce mistakes. It makes them invisible — and an invisible mistake repeats somewhere more expensive.

Same meeting, one question apart

Questions that close the room
  • "Who pushed this?"
  • "How did you not notice?"
  • "Didn't you test it?"

All three are answered with defence. Information gathered: none.

Questions that open it
  • "What were you seeing on your screen at the time?"
  • "Why did this change look safe to you?"
  • "How did the system let this reach production?"

All three interrogate the system. Information gathered: the actual cause.

The second column is not there to protect anyone; it is there because it produces better answers. "Let's be careful" is not an action item; "let's require a second approval for this class of change" is.

Four measurable signals

Measuring safety with a survey is slightly absurd: where there is no safety, the survey is not answered honestly either. Behaviour is more reliable.

1 — Who speaks first?

In the last five meetings, who offered the first opinion? If it is always the same two people, you do not have a team of ten; you have a team of two. Simple fix: the manager speaks last, and the round starts with the most junior person.

2 — How late does bad news arrive?

The delay is the most honest indicator there is. If "this will not be finished" shows up on the last day of the sprint, nobody believes saying it early is safe.

3 — How many people can say "I don't know"?

If you never hear it, it is not because everyone knows everything; it is because people have learned that not knowing carries a cost. The fastest way to reverse that is for the manager to say it first.

4 — What happens after a disagreement?

Recall the last time someone openly disagreed with you. What happened to that person afterwards? The team can answer this far more precisely than you can — because they were all watching.

The day the manager admits a mistake

One sprint I prioritised badly: two weeks of work shipped a feature nobody used. I opened the retro with it myself: "I made that call and I made it wrong. Here is what I skipped: we started without ever looking at usage numbers."

Nobody applauded, and there was no warm moment. The change showed up in the next two retros: people started describing their own mistakes. One "I estimated that badly", one "I pushed for that library and it did not work out". I had not heard sentences like that in two years.

The reason is simple: a team learns the cost of admitting a mistake by watching whoever carries the most risk. That is you.

The common misreading

Psychological safety does not mean lowering the bar. Quite the opposite: high standards plus low safety equals fear; high standards plus high safety equals learning. Avoiding hard feedback does not create safety, it just produces a team that does not talk — which is what negative feedback is for.

Checklist

A safety read on your team
  • In the last five meetings, was the first idea always from the same people?
  • How many days late did the last piece of bad news reach me?
  • How often did I say "I don't know" this month? And the team?
  • What question did I open the last incident review with?
  • Who last disagreed with me openly, and when?
  • When did I last admit a mistake in front of the team?
  • Did I follow up with the quiet person — after the meeting, not in it?

Conclusion

Psychological safety is not a feeling; it is a learned expectation: if I say the true thing here, nothing bad happens to me. That expectation is built by repeated behaviour rather than statements, and a single bad meeting can undo months of it.

The good news is that it is measurable. Look at how many days it takes bad news to reach you — that number is the culture you have built.