How do we run mentorship when we’re already behind on roadmap?

Treat it like reliability work with planned capacity. Start with 2 hours/week per rotation for one critical system. The first wins are usually faster incident response and fewer “only Sam can review this” bottlenecks—both pay back time quickly.

We’re regulated (SOX/HIPAA). Can mentorship still work?

Yes, but you need to design for access controls and traceability. Pre-provision least-privilege roles, bake approval steps into runbooks, and produce artifacts via PRs/ADRs so the audit trail is automatic.

What’s the minimum viable mentorship program?

Pick one critical flow, schedule a weekly 30-minute system tour + 30-minute runbook drill for 4 weeks, and require one artifact per week (runbook PR, dashboard update, or ADR). Track onboarding time-to-first-PR and pager load spread.

How do we avoid making mentors a bottleneck?

Rotate mentors, cap mentorship load (e.g., one mentee at a time), and reduce interruption debt by protecting focus blocks. Use `CODEOWNERS` and review rules to spread review responsibility without dumping everything on the same expert.

Culture · Jan 16, 2026 · 8 minute read

The “Only Sam Knows” System: Mentorship Programs That Stop Your Bus Factor From Killing Releases

If your platform survives on folklore, you don’t have “tribal knowledge” — you have a single point of failure. Here’s how to build a mentorship program that transfers critical system knowledge with rituals, artifacts, and metrics that work in real enterprises.

Back to all posts

The “Only Sam Knows” System: Mentorship Programs That Stop Your Bus Factor From Killing Releases

Key takeaways

Implementation checklist