Psychological Safety

Scenario

Example team: in the payments team, a junior engineer noticed a suspicious rounding behaviour in refunds but didn't mention it — the last time someone raised a concern in planning, the tech lead dismissed it in front of everyone. Three weeks later the bug cost the company a reconciliation incident.

What you see on the team

• Few questions in meetings; silence when the lead asks "any concerns?".
• Mistakes discovered late, often by someone outside the team.
• Juniors who never challenge seniors.
• Postmortems where people are defensive.

The common mistake

Assuming that telling people "you can always speak up" is enough, while day-to-day reactions to mistakes and questions teach the opposite.

Why it happens

People learn what's safe from how leaders react, not from what they say. One dismissive comment can outweigh many invitations. Google's Project Aristotle studied 180 teams and found that what mattered was "less about who is on the team, and more about how the team worked together" — with psychological safety first among the five dynamics, followed by dependability, structure and clarity, meaning, and impact (re:Work).

The playbook

Model fallibility. Admit your own mistakes and uncertainty openly: "I got this estimate wrong", "I don't know — who does?".

React well to bad news. Thank people for raising problems, especially early and especially when it's uncomfortable. Your reaction in that moment is the whole lesson.

Invite input explicitly and structurally:

• ask the most junior people first in discussions;
• silent writing before discussion;
• "what could go wrong with this?" as a standard design-review question.

Separate learning from blame: blameless postmortems; treat honest mistakes as information.

Address behaviour that undermines safety quickly and privately — dismissive comments, interrupting, sarcasm — regardless of seniority.

Measure it: a short anonymous pulse every quarter, e.g. "If I make a mistake on this team, it is not held against me" (one of the survey items used in Google's research).

What to say

When someone raises a concern late:

"Thank you for raising this — I'd much rather hear it now than after release. Let's look at it together."

Opening a design discussion:

"Before I share my view, I'd like to hear the risks you see. Lee, you've been closest to the refund code — what worries you?"

Privately to a senior engineer who dismissed a junior's question:

"In planning, when Ana asked about the rounding rule, you said 'that's obviously fine' and moved on. She hasn't spoken up since. Her question turned out to be right. Can we make sure people feel safe asking, even when the answer seems obvious to us?"

Trade-offs

• Safety isn't comfort: high-safety teams disagree more openly, not less. Safety means you can challenge without fear.
• Safety without accountability becomes complacency; keep standards high and make it safe to fall short and learn.
• It takes time to rebuild after damage; consistency over months matters more than one gesture.

Signals it's working

• People raise risks and mistakes early, including juniors.
• Meetings include real disagreement that stays respectful.
• Postmortems surface contributing factors freely.
• Pulse survey scores on safety rise over quarters.

Key points

• Google's research found psychological safety the most important of five team dynamics.
• People learn safety from how leaders react to mistakes and bad news — not from slogans.
• Model fallibility, thank people for early bad news, and invite input structurally.
• Address undermining behaviour quickly; safety and high standards go together.