You can't create psychological safety in the middle of a crisis. The engineer who calls into an incident bridge to say "this was my change" only does that if they've already learned the organization won't punish them. That learning happens through how you handle everyday mistakes and uncertainties.
Example
When a Google engineer realized their canary deployment had caused a system-wide outage, they didn't hide. They called into the escalation bridge, identified their change, and started rolling it back. This happened because Google had built a culture where small failures were treated as learning opportunities, where missed deadlines could be discussed openly, and where asking questions—even ones that seemed obvious—was explicitly encouraged and modeled by senior people.
Context
Watch for the quiet indicators of fear: calls going silent when someone uses an unfamiliar acronym, juniors talking significantly less than seniors, people refusing to revisit old decisions because "that was decided years ago and I don't know why." These signal that people don't feel safe admitting uncertainty. The practice of continuous retrospectives for incidents of all sizes, not just the high-profile ones, reinforces that learning matters more than blame. As one leader put it: why would you fire someone who just gained thousands of dollars worth of learning?
When This Might Not Apply
– Regulated industries where legal/compliance requirements mandate individual accountability for certain failures. – Toxic cultures where leadership actively undermines safety signals (requires leadership change first).
Variations
– For remote teams, use async retrospectives to normalize small failure discussions. – In cross-team incidents, explicitly invite "silent contributors" to share observations post-resolution.
