A director closed her laptop after a steering call...
Her calendar was full. Her voice wasn’t.
Another great Harvard Business Review research found the lowest psychological safety sits with middle managers. Not with executives or frontline teams. Actually, those folks in the middle.
So, what's the real problem?
We treat psychological safety as a “team dynamic.” Yet the biggest risk lives right in the “middle layer.” Managers carry pressure from above and doubt from below. They own delivery, but rarely own the rules. In AI programs, that squeeze feels and is much tighter. New tools shift power and workflows.
When managers don’t feel safe, they slow or silence change, quietly.
What to do:
Give explicit air cover: state, in writing (it makes it clear), what managers can test, break, and question. Give freedom. Name it and model “failing forward.”
Make pilots manager-designed: ask managers to pick one workflow to improve this month. Then score effort vs. impact with their teams. Track one week-one metric they choose. Compare and analyze it, together.
Run manager circles: small groups. Same roles. Clear norms. Share what failed and what they tried next. Capture patterns and adjust playbooks. Co-create.
Measure safety where work happens: use short pulse items tied to moments: standups, reviews, approvals.
Coach the middle like a product: map pain points. Ship tiny fixes: office hours, templates, checklists. Celebrate a single “win the week” story, every week.
If you want adoption to succeed, protect your middle managers and be their biggest supporter.
Sources: Harvard Business Review
