Psychological Safety

Theory & origin

Amy Edmondson coined team psychological safety in a 1999 study with a famous twist: the better hospital teams appeared to make more medication errors, until the data showed they were simply willing to report them. The worse teams were hiding theirs. Google's Project Aristotle later found psychological safety the strongest predictor of team effectiveness across hundreds of its own teams, ahead of composition and individual talent. Timothy Clark's four-stages ladder turned the concept into a usable staircase: inclusion safety, learner safety, contributor safety, and challenger safety, climbed in that order. The model's sharpest edge is diagnostic: a team stalls exactly one step below where its leader is comfortable being challenged. And its most common corruption is equating safety with comfort. Edmondson pairs it with accountability: high safety plus high standards is the learning zone, high safety alone is just a pleasant place to underperform.

Key components

The parts at a glance. Click any term for the full definition, a field example, and the common failure, in the model below.

How a consultant runs it

1. 01 Measure the stages separately. An average hides the pattern, and the pattern is the diagnosis: most teams score high on inclusion and collapse at challenger.
2. 02 Read silence as data. Near-zero error reports, no questions in reviews, and unanimous decisions are the sound of fear, not excellence.
3. 03 Work on leader behavior, not team posters. Safety is set by what happens to the next person who admits a mistake or challenges the boss.
4. 04 Have leaders go first: publicly owning what they got wrong this month buys more safety than any workshop.
5. 05 Pair every safety intervention with standards. The goal is the learning zone, candor plus accountability, not comfort.

Worked example

Twelve operations squads at Bank Nusantara show a strange pattern: one branch cluster reports almost no incidents while processing the same volume as everyone else. The staged survey explains it: inclusion 4.1, learner 3.4, contributor 2.9, challenger 1.8. People are welcome, allowed to ask, mostly heard, and completely unwilling to challenge. The silence was being read as excellence. The intervention targets leader behavior: blameless post-incident reviews, cluster heads opening monthly meetings with what they themselves got wrong, and public thanks for the first challengers. Within a quarter, error reports triple, which is the good version of that sentence, repeat incidents halve because problems now surface while they are small, and tellers flag two near-miss fraud patterns nobody upstairs had seen.

Common pitfalls

1. 01 Confusing safety with comfort, and building a nice team that never tells the truth
2. 02 Running a survey and skipping the leader-behavior work, which is where safety is actually set
3. 03 Reading low error reports as good performance instead of as possible silence
4. 04 Punishing one visible challenger, which resets years of accumulated safety in an afternoon
5. 05 Treating it as an HR program instead of a property of how each leader responds to bad news

Sample deliverable

One real engagement, start to finish. Watch the numbers travel from raw input, onto the chart, into the finished artifact.