Protocol

Let a colleague rehearse the runbook without your missing context

The missing step is usually obvious to the person who forgot to write it.

When it fits

  • A procedure looks complete to its author but must work for another qualified operator.

When to avoid it

  • Never create uncontrolled production failures to test documentation. A runbook cannot replace prerequisite training, access or judgment.

Why it matters

Give a colleague a safe practice scenario and the documented starting conditions. Observe where they need information that exists only in your head. Use the rehearsal to repair the procedure, including expected results and escalation points, rather than grading the colleague for not reading your mind.

Steps

  1. Choose an isolated or simulated case with a safe reset and clear stop conditions.
  2. Let the colleague follow the documented route; record needed prompts and ambiguous steps.
  3. Revise the runbook and repeat the affected part with the agreed support level.

An example

A new support engineer rehearses a failed import. The exercise reveals that the runbook never identifies which log belongs to the processing attempt.

Check your result

The colleague can reach the expected result or correctly escalate using the stated prerequisites and support.

Keep this limit in mind

  • Never create uncontrolled production failures to test documentation. A runbook cannot replace prerequisite training, access or judgment.

Evidence and sources

Supports

The SRE on-call account uses practical exercises and incident role-play alongside documented playbooks, including semi-independent onboarding exercises.

Testing a runbook with a colleague is an editorial extension; the source does not isolate its causal effect on incident outcomes.

Being On-Call · Training roadmap; Afterword; Maintaining Playbooks

All sources (1)