Psychological Safety: How to Build It as an Engineering Manager

2026-07-10 · 7 min read · Team Health

In 2012, Google ran a large internal study called Project Aristotle, trying to understand what made some engineering teams consistently outperform others. They tracked dozens of variables: individual talent, years of experience, seniority mix, personality types, work styles. None of them explained performance differences reliably.

The one factor that did: psychological safety. Teams where members felt safe to take interpersonal risks — to speak up, disagree, ask questions, admit mistakes — consistently outperformed those where they didn't.

This finding has been replicated widely. And yet psychological safety remains one of the most misunderstood concepts in management.

What Psychological Safety Actually Is

Psychological safety is not about being nice. It's not about avoiding conflict. It's not about making people feel good.

Amy Edmondson, the Harvard researcher who developed the concept, defines it as "a belief that the team is safe for interpersonal risk-taking." The operative word is risk. Can people say things that might be wrong? Can they challenge decisions made by someone senior? Can they admit they don't understand something without feeling judged? Can they raise a concern without it being used against them?

A team with high psychological safety has real disagreements, honest critique, and people who challenge each other's ideas. It can feel uncomfortable. What it doesn't have is the fear of social punishment for speaking up.

This distinction matters because managers sometimes confuse psychological safety with conflict avoidance. They keep things positive and smooth — no hard feedback, no challenging conversations — and wonder why nobody raises problems until they've become crises.

The Manager's Outsized Influence

Team culture is heavily shaped by the most senior person in the room. Whether you intend to or not, your behaviour sets the norms. If you respond defensively to challenge, people stop challenging. If you penalise mistakes, people hide them. If you reward the loudest voice over the most thoughtful one, quieter people go quiet.

This means the most direct lever for building psychological safety is your own behaviour, not workshops or team-building exercises.

Specifically:

How you handle being wrong. If you never admit mistakes, you signal that admitting mistakes is dangerous. Engineers will cover theirs up, attribute problems to external factors, and avoid sharing early warnings about issues. Model the behaviour you want: "I was wrong about the timeline estimate on this. Here's what I missed."

How you respond to bad news. When someone brings you a problem, your reaction in that moment teaches the rest of the team what happens when you bring problems. Shoot the messenger once and watch problem reporting drop to zero. Respond with "thanks for flagging this early, let's figure it out together" and you get more early warnings.

How you handle dissent. If someone challenges your technical decision and you overrule them without engaging the reasoning, you signal that challenge is futile. If you engage seriously — explaining your thinking, acknowledging their point, sometimes changing your mind — you signal that challenge is worth doing.

How you talk about failure. Blameless postmortems are standard practice in high-safety teams because they decouple incident analysis from individual punishment. If postmortems are actually blame sessions with a different name, the psychological safety effect disappears.

Practical Techniques

Ask for input, especially from the quiet ones. In meetings, actively solicit views from people who haven't spoken. "Priya, you know this codebase well — what do you see that we're missing?" This signals that silence is not the norm and that all perspectives are sought.

Reward the raising of problems, not just the solving. When someone flags a risk or raises a concern, acknowledge it specifically. "Thanks for raising this — this is exactly the kind of thing we need to surface early." Make the act of speaking up visibly valued.

Create low-stakes ways to be wrong. Not every statement needs to be a commitment. In design discussions, encourage speculation: "What if we tried X?" "What's the worst case if we're wrong here?" Hypothesis-framing reduces the perceived risk of floating ideas that might not work.

Debrief failures without extracting blame. When something goes wrong, the first conversation should be about what happened, not who caused it. Systems thinking over individual blame. "What conditions allowed this to happen?" is more psychologically safe than "who approved this change?"

Separate idea critique from personal critique. "This approach has a problem I want to flag" is safe. "You clearly didn't think this through" is not. This sounds obvious but in high-pressure moments the line blurs quickly. Be intentional.

Measuring Psychological Safety

You can get a rough read on your team's psychological safety through a few questions asked anonymously:

If the answers cluster around "sometimes" or "no," you have work to do. If they cluster around "yes," maintain it — psychological safety is easier to lose than to build.

The Long Game

Building psychological safety is a long process and a fragile state. A team that took a year to reach high safety can lose it in a month if the environment changes — a new manager who responds poorly to dissent, a period of high stress that reduces tolerance for mistakes, a high-profile blame response to a major incident.

This means it's not something you build and walk away from. It requires consistent maintenance: noticing when the signals of safety are declining (people getting quieter, problems going unreported, postmortems becoming defensive) and actively resetting.

The teams that sustain it over time don't think of psychological safety as a culture initiative. They think of it as a prerequisite for doing their jobs well — because it is.