Every Friday at 5pm, one question decides whether your weekend is quiet or ruined: "If the on-call person disappeared right now, could someone else run the week?"
If the honest answer is "no" — you don't have a team problem, you have a runbook gap. Here's the 10-minute audit that finds the gaps before the pager does.
The Monday Morning Audit (10 minutes)
1. The Stranger Test. Pick your most critical recurring task. Ask: could a competent stranger do it with only what's written down? If the answer lives in someone's head, that's gap #1.
2. The Single-Name Scan. List your critical systems. Next to each, write who owns it. Any system with ONE name next to it is a bus-factor outage waiting to happen — not hypothetical, statistical.
3. The Last-Fire Replay. Take the last 3 incidents. For each: was there a written fix, or a Slack scavenger hunt? Every "we figured it out" without a runbook is a fire you've decided to have again.
4. The Restore Drill. Backups nobody has restored are wishes, not backups. Restore one thing — anything — today. The first restore always takes 4x longer than planned. Better to learn that on a Tuesday.
5. The Handoff Check. Look at yesterday's shift handoff. Was it a written note or a hallway conversation? Hallways don't work across time zones, sick days, or resignations.
What a runbook actually needs
Not documentation. Documentation is a novel; a runbook is a map. One page per critical task:
- Trigger: what page/alert starts this
- Fix: the exact commands/steps, copy-pasteable
- Verify: how you KNOW it's fixed (not "it seems better")
- Escalate: the name and number for when step 2 fails
If you can't copy-paste the fix, you don't have a runbook — you have a vibe.
Start with the money systems
Audit in this order: anything that bills customers, anything that takes payments, anything that ships product. A broken blog is an embarrassment; broken billing is a bankruptcy. The order matters because audit time is scarce and the blast radius isn't equal.
I turned this audit into a fill-in-the-blanks kit — the stranger test checklist, a bus-factor matrix template, and incident replay forms — in the Ops Starter Kit. If you're running a one-person or small ops team, the Automation Starter Pack covers the automation side of the same problem.
What's the one system at your work that only ONE person understands? Drop it in the comments — I'll send back the runbook skeleton for it.
Top comments (0)