Capturing institutional knowledge: a manager's checklist

Use this manager's checklist to capture institutional knowledge before attrition, role changes, or reorgs — structured prompts for pre-departure sprints, role-critical audits, and quarterly reviews.

This supports the full framework in how to document tribal knowledge before it walks out the door. Print it, run it in 60-minute sessions, and store outputs where ops — not just HR — can use them.

Pre-departure checklist (notice received)

Complete within the first week of notice period:

• List the 3 workflows only this person fully owns end-to-end
• Identify downstream roles that depend on their routing decisions
• Document system logins and access tied to tribal fixes (shared accounts, workaround tools)
• Capture client/program exceptions they apply by memory
• Record external contacts (brokers, vendors, regulators) they manage informally
• Schedule shadow sessions — successor watches exception handling, not just happy path
• Run interview question bank below (session 1: daily workflow; session 2: exceptions)
• Assign interim router with written routing rules extracted from interviews
• Store artifacts in ops-owned location (not personal drive)

Red flag: If they say "it's all in my head," escalate to structured discovery — two weeks is not enough for open-ended transfer.

Role-critical audit (quarterly)

For any role where throughput drops when one person is out:

• Confirm backup coverage exists and has been tested in last 90 days
• Update routing rules written from last quarter's interviews
• List new exceptions added since last audit (clients, programs, tools)
• Check shadow tools (spreadsheets, trackers) — still accurate?
• Interview downstream consumers — do they still re-verify this role's output?
• Score automation potential — eliminate step vs document step vs bot step
• Link findings to automation roadmap backlog

New hire / onboarding failure checklist

Run when ramp exceeds 90 days or error rates stay elevated:

• Is tribal knowledge the bottleneck (undocumented) vs training design?
• Which steps require asking a specific person vs reading SOP?
• Where does the official SOP diverge from actual practice?
• Can a new hire complete one transaction without interrupting a senior employee?
• If no — capture gaps using interview bank below

Interview question bank

Use open probes; follow up for specifics (minutes, systems, frequency).

Daily workflow

1. Walk me through yesterday's work — where did you start and end?
2. Which steps are not in the official SOP?
3. Where do you copy data between systems manually?

Exceptions 4. When does the standard process not apply? 5. What do you do when [common failure from Q4] happens? 6. Who do you escalate to — and what do they do that you cannot?

Handoffs 7. Who depends on your output before they can start? 8. What do they complain about or re-check? 9. Where does work wait on you with no visibility for them?

Elimination 10. If you could remove one step entirely, which would save the most time? 11. What would break if you were out two weeks?

Record answers with role, date, and interviewer. Cross-validate with upstream/downstream — see finding hidden bottlenecks.

Storage and maintenance rules

• Ops owns the repository — HR exit files are insufficient for workflow detail
• Version and date every capture session
• No personal drives — use shared ops wiki or knowledge base with search
• Review quarterly — stale capture is worse than none (false confidence)
• Link to roadmap — each recurring pain becomes a ranked automation candidate

When checklists are not enough

Checklists excel for critical single roles and urgent departures. They do not scale to:

• 50+ employees across functions
• Cross-validating the same bottleneck from multiple angles
• Producing a deduplicated, ROI-ranked org-wide map

For that, use parallel structured interviews — operational intelligence at workforce scale. The checklist questions become the interview script; the synthesis layer automates.

Next action

Identify one hero dependency on your team. Schedule a 60-minute session using questions 1–6 this week. Store output in an ops-owned folder. If downstream roles validate the same pain, add it to your automation backlog with evidence.