💡 Core idea
Trust is not a soft extra. It is what makes speed, candor, and delegation possible. It is built in small, consistent acts and lost in public moments of carelessness.
You cannot delegate to someone who expects to be undercut. You cannot hear risks from someone who was punished for raising the last one. Speed that depends on silence is not speed. It is delay with a smile.
Amy Edmondson (1999), in a study of 51 work teams, defined team psychological safety as a shared belief that the team is safe for interpersonal risk taking. That climate was associated with learning behavior — asking for help, talking about errors, seeking feedback — and learning behavior mediated the link to team performance. The mechanism is the one this chapter is about: people stay silent when they expect to be undercut.
Ben Kuhn treats trust as a bottleneck to growing a team: when it is missing, everything else gets more expensive. Build it on purpose.
Competence, reliability, and care
Three pillars you can actually manage:
Competence. You know the work well enough to set a standard, to help, and not to be fooled. You do not have to be the best individual contributor. You do have to be a serious counterpart.
Reliability. You do what you said, when you said it, or you renegotiate early. A commitment log is more useful than a speech about integrity.
Care. People are not instruments. You can be demanding and still make it obvious that you will not trade their dignity for your convenience.
Stay SaaSy puts a public test on this: in the room, a manager stays on the team’s side. That is honor, not theater. Managing up (Chapter 16) does not include abandoning your people to look aligned.
Kore Nordmann belongs here as well as in Chapter 1. Honesty about hierarchy is part of trust. Pretending you are simply a friend papers over a power imbalance everyone can already see.
How trust is broken — especially by surprise
Trust is lost less often on the failure itself than on the surprise. Ben Balter’s rule is to tell people now if the news would surprise them later. Being “on track” is not the same as keeping them in the loop.
The same wound has other shapes: private criticism made public, shifting goals with no explanation, taking credit, hiding bad news until there is no time to act. Each one teaches the team to stop bringing you reality.
After a miss, do not stop at understanding. Michael Heap separates “I now understand the incident” from “we have changed the system so this class of failure is less likely.” Ask what will change. A thorough post-mortem that changes nothing is another form of surprise: you performed seriousness and then repeated the pattern.
Repair
You will break trust. The question is whether you can repair it.
Name what happened, in specific terms. Take responsibility without a tour of your intentions. Change a visible behavior — not a mood. Tell the other person how they can hold you to it. Do not demand instant forgiveness. Repair is a sequence of acts, not a single apology.
Transparency is not oversharing
People need timely truth about reorgs, performance, and any tool that watches their work. They do not need your unfiltered anxiety, or information you do not have the right to share. The test is practical: would withholding this now create a surprise later that they could have used? If yes, speak. If you are using the conversation to make yourself feel honest, wait and find a more careful form.
Tools
- No-surprise rule. Important news comes from you first, in time to respond — downward and upward.
- Commitment log. Promises made in 1:1s and team meetings, reviewed weekly. Unkept promises are how reliability quietly dies.
- Trust-repair script. What I did; what I understand it caused; what I will do differently; how you can hold me to it.
- Incident question. After a failure: what will we change so this class of event is less likely?
🤖 AI-era tip: Do not outsource the human truth of a relationship
You may use an approved tool to structure a difficult conversation or to check that a written note is specific rather than vague. You must not paste sensitive personal or personnel details into an unapproved tool, and you must not send machine-drafted feedback you have not fully rewritten in your own voice. If a tool helped you prepare in a way that would matter to the other person, be ready to say so. Trust dies when people feel they are talking to a generator, or being quietly scored by one.
Takeaways
- Trust rests on competence, reliability, and care — consistency beats intensity.
- Surprises destroy more trust than honest failures; say it now if it would surprise them later.
- After an incident, the useful question is what will change.
- Repair is a visible change in behavior, not a demand for closure.
- Stay on your team’s side in the room, and be honest about hierarchy.
Evidence
- Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383. In 51 manufacturing work teams, team psychological safety was associated with learning behavior; learning behavior mediated the relationship with team performance. Safety is not group cohesiveness: cohesiveness can suppress challenge (groupthink).