Vanity metric always grows and unaffected by decisions — harmful because feels like success.

Every 6 months with each report, ask: what keeps you here? What might make you leave?

Every writing = one CTA. Two = zero.

All data: public/internal/confidential/restricted — each class with different protection rules.

Ask every team member once a quarter: where do you want to be in two years? Then explicitly tie their current work to that path.

Every new member gets a peer "buddy" (not their manager) for the first 30 days to ask any "silly" question.

30 min with a top performer, 5 questions on "why you stay and why you might leave" before they actually consider leaving.

Ask the model to question you instead of answering, surfacing and deepening your understanding.

After a Socratic learning session, ask AI to generate retrieval questions at increasing intervals (1 day, 3 days, a week) instead of relying on random memory.

An audience watches you silently; a community interacts with each other even without you — assess which you're actually building and which serves your goal.