Post

Building Trust & Psychological Safety

Psychological safety is not about being nice -- it is about making it safe to take interpersonal risk. High-performing teams are not conflict-free teams. They are teams where conflict is productive because people trust that disagreement will not be punished.

Posted Updated
By
14 min read
Building Trust & Psychological Safety

Amy Edmondson’s Research — What the Data Actually Shows

Amy Edmondson’s research at Harvard (1999, expanded in “The Fearless Organization,” 2018) is the most rigorous work on psychological safety. The findings are counterintuitive and frequently misunderstood.

The Core Finding

Edmondson’s original study at hospitals found that the best-performing medical teams reported MORE mistakes, not fewer. The teams that appeared to have fewer errors were not actually making fewer errors — they were hiding them. The high-performing teams had a culture where reporting errors was safe, which meant errors were caught and corrected faster.

This maps directly to engineering: teams with low incident counts might not have fewer incidents — they might have a culture where people hide failures, work around bugs instead of reporting them, and avoid risky improvements. The healthiest engineering teams have well-documented incident histories because they surface problems openly.

What Psychological Safety IS and IS NOT

Psychological Safety IS Psychological Safety IS NOT
Confidence that the team will not embarrass, reject, or punish you for speaking up Niceness, comfort, or absence of conflict
A belief that it is safe to take interpersonal risk (asking questions, admitting mistakes, challenging ideas) A belief that everything you say will be agreed with
A team-level phenomenon (not individual personality) Being “soft” or lowering standards
A prerequisite for high performance A substitute for accountability
Compatible with high standards and direct feedback A reason to avoid giving hard feedback

The Performance Matrix

Edmondson’s 2x2 maps psychological safety against accountability/standards:

  Low Standards High Standards
High Psychological Safety Comfort Zone — People feel safe but no one pushes for excellence. Nice but mediocre. Learning Zone — The target. People challenge each other, take risks, and hold each other accountable. High performance.
Low Psychological Safety Apathy Zone — No one cares and no one speaks up. Worst case. Anxiety Zone — High pressure without safety. People are afraid to take risks or admit mistakes. Burnout, hiding failures, CYA behavior.

The key insight: Psychological safety without high standards produces comfort, not performance. You need BOTH. The leader’s job is to hold the tension — make it safe to take risks AND demand excellent outcomes.

The Trust Equation (Maister, Green & Galford, 2000)

From “The Trusted Advisor,” a practical model for understanding what builds and erodes trust:

1
2
3

                Credibility + Reliability + Intimacy
Trustworthiness = ────────────────────────────────────────
                         Self-Orientation
Component Definition How to Build It How to Destroy It
Credibility Expertise and accuracy. “I trust their words.” Be right. Admit when you are wrong. Do not bluff. Bullshit. Overstate your knowledge. Give advice outside your competence.
Reliability Consistency and follow-through. “I trust their actions.” Do what you say. Show up on time. Deliver when promised. Small commitments kept matter more than grand promises. Miss deadlines. Forget commitments. “I will get back to you” and never do.
Intimacy Emotional safety. “I trust them with my concerns.” Listen without judging. Share your own struggles. Keep confidences. Gossip. Share 1:1 conversations with others. React punitively to vulnerability.
Self-Orientation (denominator) Focus on self vs. focus on others. The most powerful lever. Ask questions before giving answers. Show genuine curiosity about the other person’s situation. Talk about yourself. Make their problem about you. Hijack conversations. Use information for personal advantage.

Why self-orientation is the denominator: It is the most destructive factor because it multiplies distrust. An EM who is highly credible, reliable, and intimate BUT primarily focused on their own career advancement is trusted less than a moderately credible, reliable leader who genuinely prioritizes the team’s success.

Applying the Trust Equation in Engineering Management

With your team:

  • Credibility: Know your domain. If you are not technical enough to understand the problems, say so and defer to those who are. Pretending to understand erodes trust faster than admitting you do not.
  • Reliability: If you promise to escalate a blocker, escalate it today. If you promise to follow up on a career growth conversation, follow up this week. The speed of your follow-through is a trust signal.
  • Intimacy: Remember personal details. Know that one engineer is going through a divorce and another is training for a marathon. Not to exploit — to care. When they share struggles, do not immediately problem-solve. Listen first.
  • Self-Orientation: In public forums, give credit to the team. In private conversations, take blame on yourself. “The team did great work on X” and “I should have caught that risk earlier.”

Stephen M.R. Covey’s Speed of Trust (2006)

Covey’s framework adds an economic lens: trust is not just nice to have — it directly affects speed and cost.

The Trust Tax and Trust Dividend

1
2

High Trust = High Speed + Low Cost     (Trust Dividend)
Low Trust  = Low Speed  + High Cost    (Trust Tax)

Engineering examples:

Scenario High Trust Low Trust
Code review Quick review, constructive feedback, approved same day Multi-day review, defensive comments, escalation to architect
Incident response Open postmortem, root cause found quickly, no blame Cover-up, finger-pointing, same incident recurs
Cross-team dependency Informal agreement, handshake, delivered on time Formal contract, SLA negotiation, weekly status meetings, still delivered late
New initiative proposal EM proposes to VP, gets approval in one meeting EM prepares 40-slide deck, presents to committee, three rounds of revision

The trust tax in engineering organizations is enormous. Every process, approval gate, and review committee that exists because people do not trust each other is a direct cost. Warren Buffett famously closed the $23 billion acquisition of McLane from Walmart with a handshake and no due diligence — the trust dividend in action.

The 13 Behaviors of High-Trust Leaders

Covey identifies 13 behaviors. The five most relevant for engineering leaders:

  1. Talk straight. Say what you mean. Do not hedge, sugarcoat, or use corporate-speak. Engineers especially value directness. “This project is at risk” is more trustworthy than “we have some challenges but I am confident we will find a path forward.”

  2. Create transparency. Share information proactively. Default to open. When you cannot share something, say “I cannot share the details yet, but here is what I can tell you.” Transparency gaps get filled with rumors.

  3. Right wrongs. When you make a mistake — and you will — apologize specifically and quickly. “I should not have made that decision without consulting the team. I was wrong. Here is what I will do differently.” The longer you wait, the harder it gets and the more trust erodes.

  4. Show loyalty. Never talk negatively about absent team members. If someone criticizes your team member in a meeting, defend them (or commit to addressing it privately if the criticism has merit). Your team is watching how you handle this.

  5. Deliver results. Trust without results is wishful thinking. The fastest way to build trust is to consistently deliver on commitments. All the relationship-building in the world does not compensate for a team that chronically misses targets.

Vulnerability-Based Trust (Lencioni, 2002)

Patrick Lencioni’s “The Five Dysfunctions of a Team” places trust at the foundation of the pyramid:

1
2
3
4
5

5. Inattention to Results
4. Avoidance of Accountability
3. Lack of Commitment
2. Fear of Conflict
1. Absence of Trust        ← Foundation

Each layer depends on the one below it:

  • Without trust, people fear conflict (they will not challenge each other)
  • Without healthy conflict, people do not commit (they nod but do not buy in)
  • Without commitment, people avoid accountability (why hold someone to something they never agreed to?)
  • Without accountability, people do not focus on results (individual goals override team goals)

Vulnerability-Based Trust in Practice

Lencioni’s trust is specifically vulnerability-based: the willingness to admit weaknesses, ask for help, acknowledge mistakes, and say “I do not know.”

Why this is hard for engineering leaders: Technical cultures reward competence. Admitting you do not know something feels like a career risk. But paradoxically, leaders who pretend to know everything are trusted less than leaders who openly acknowledge their gaps and surround themselves with people who fill them.

The leader goes first. You cannot ask your team to be vulnerable if you are not. Practical demonstrations:

  • In a team meeting: “I made a mistake on the priority call last sprint. Here is what I should have done differently.”
  • In a 1:1: “I am still figuring out how to best support the AI platform work. What am I missing?”
  • In a retrospective: “What is something I could do better as a manager?”

The response matters more than the question. If an engineer admits a mistake and you react with frustration or assign blame, you have taught the entire team that vulnerability is punished. One bad reaction undoes months of trust-building.

Building Psychological Safety — Practical Actions

For Team Meetings

  1. React to mistakes with curiosity, not blame. When an incident happens: “What can we learn from this?” not “Who caused this?”
  2. Explicitly invite dissent. “I have shared my thinking. What am I missing? Where is this wrong?” And then wait. Silence is uncomfortable — let it sit.
  3. Celebrate learning from failure. Share your own failures and what you learned. Normalize that learning requires failing.
  4. Call out interruptions. If someone is cut off: “Hold on — [name] was making a point. Let them finish.”
  5. Respond to questions with respect, regardless of the question. If a junior engineer asks a basic question, answer it fully and thank them for asking. The team is watching.

For 1:1s

  1. Ask “what is something that is hard to talk about?” regularly. Not every time — but often enough that it is normal.
  2. When someone shares something difficult, listen fully before responding. Do not immediately jump to solutions.
  3. Follow up on previous concerns. “Last time you mentioned [X]. How is that going?” This signals that you were actually listening.
  4. Share your own challenges. “I am struggling with [X]. What do you think?” This models vulnerability.

For Incident Response

  1. Blameless postmortems — but for real. The word “blameless” has become a checkbox. Actual blameless postmortems require: focusing on systemic causes (not individual actions), assuming good intent, asking “what about our system allowed this to happen?” instead of “who did this?”
  2. Remove the phrase “human error” from your vocabulary. If a human can make an error, the system should prevent or catch it. The system failed, not the person.
  3. Protect the people involved. If leadership demands someone to blame, push back. “The postmortem found systemic issues. Blaming an individual will ensure these issues are never reported again.”

Trust Repair — When It Breaks

Trust breaks. People make mistakes, promises get broken, miscommunications happen. The quality of the repair matters more than the break itself.

The Anatomy of Trust Repair

Step 1: Acknowledge the break specifically. Not “I am sorry if anyone was hurt” but “I promised I would advocate for the team’s technical decision in the leadership meeting, and instead I agreed to the VP’s approach without pushing back. I broke my commitment to you.”

Step 2: Own the impact without justifying. Do not explain why you did what you did (yet). First, acknowledge the impact. “I understand that this made you feel like I do not have your back.”

Step 3: Explain (briefly) without excusing. “I was caught off guard in the meeting and did not handle the pressure well. That is not an excuse — it is a gap I need to work on.”

Step 4: Commit to specific behavioral change. Not “I will do better” but “Next time, I will ask to table the discussion so I can consult the team before committing. And I will share leadership meeting context with you proactively so you are not surprised.”

Step 5: Follow through visibly. The repair is not the conversation — it is the sustained changed behavior. One conversation rebuilds credibility; six months of consistent new behavior rebuilds trust.

When Trust Is Irreparably Broken

Sometimes trust cannot be repaired:

  • Repeated breaches after stated commitment to change
  • Fundamental values misalignment (e.g., an EM who does not believe in psychological safety)
  • Betrayal of confidence that caused material harm

In these cases, the honest path is to acknowledge the break and find a structural solution (different reporting line, different team, or separation).

Measuring Psychological Safety

You cannot manage what you do not measure. But you also cannot survey your way to psychological safety if the surveys themselves are not safe.

Edmondson’s 7-Item Scale

The original research instrument (adapted for engineering):

  1. If you make a mistake on this team, it is often held against you. (reverse scored)
  2. Members of this team are able to bring up problems and tough issues.
  3. People on this team sometimes reject others for being different. (reverse scored)
  4. It is safe to take a risk on this team.
  5. It is difficult to ask other members of this team for help. (reverse scored)
  6. No one on this team would deliberately act in a way that undermines my efforts.
  7. Working with members of this team, my unique skills and talents are valued and utilized.

Behavioral Indicators (More Reliable Than Surveys)

Surveys measure perception. Behavior reveals reality.

Signal of High Safety Signal of Low Safety
People ask “why?” in meetings without preamble Questions are prefaced with “this might be a stupid question, but…”
Postmortems discuss what happened openly Postmortems are carefully worded to avoid naming anyone
Junior engineers challenge senior engineers’ ideas Junior engineers only speak when directly asked
“I do not know” is said out loud without apology People bluff or deflect rather than admitting knowledge gaps
Failed experiments are discussed as learning Failed experiments are quietly buried
Engineers flag risks early (“I think we might miss the deadline”) Engineers work overtime silently and announce delays at the last minute
Feedback flows in all directions Feedback only flows downward

Anti-Patterns

Anti-Pattern Description Why It Is Harmful
Faux Safety Leader says “this is a safe space” but punishes dissent Worse than no safety claim — it adds betrayal
Safety without standards Everything is accepted, nothing is challenged Team stagnates. “Safety” becomes an excuse for mediocrity
Confusing trust with agreement “You should trust me” means “stop disagreeing with me” Destroys both trust and psychological safety
The Vulnerability Olympics Leader shares too much, too often, creating discomfort Vulnerability is a tool, not a personality trait. Calibrate to context.
Selective safety Safe for the inner circle, unsafe for others Creates in-group/out-group dynamics. Fragments the team.
Premature trust Assuming trust exists because the team is new Trust must be earned through repeated interactions. Assume it does not exist and build it deliberately.
Trust through abdication “I trust you completely” used to avoid involvement Absence is not trust. Trust includes follow-up, support, and calibrated oversight.

References

Engineering Leadership
trust psychological-safety culture team-dynamics
This post is licensed under CC BY 4.0 by the author.