Five Whys
Five Whys. The Five Whys technique asks “why?” repeatedly to trace a problem from its visible symptom to a fixable root cause.
Summary
Five Whys looks simple on paper. In practice each answer can fork into multiple valid next questions. You are building a tree of cause chains, not a single straight line.
| Pitfall | What goes wrong | Fix |
|---|---|---|
| Literal five | Stopping at an arbitrary depth | Follow each branch until you reach a changeable process |
| Weak early whys | Filler causes that cannot be implemented | Require quality answers for the first three levels |
| Single branch only | Missing parallel causes | Start from Fishbone diagram, then Five Whys on each rib |
| Blame stopping point | “Human error” as the root | Push to process, guardrail, or automation gaps |
Facilitation rule: Tell the group the first three “why” answers must be specific and actionable. Weak early layers poison every branch below them.
Sibling technique: First Principles is the forward-looking design counterpart. Use it when choosing architecture, not when tracing an incident backward.
Details
Branching Workflow
- State the observable problem as a fact the group agrees on.
- Ask “why?” and capture every plausible answer, not just the first.
- For each answer, ask “why?” again. Track branches separately (whiteboard, doc outline, or Fishbone diagram).
- Stop a branch when the fix is a durable process, guardrail, or automation change.
- Prioritize action items on branches that recur across incidents.
Quality Bar for Early Layers
The first three answers should name mechanisms, not vibes. Compare:
| Weak | Strong |
|---|---|
| “Someone was careless” | “Runbook step 4 has no verification gate” |
| “Deploy was rushed” | “Change window has no rollback checklist” |
| “Monitoring failed” | “Alert routed to a queue with no on-call subscription” |
When to Pair with Fishbone
Use Fishbone diagram (or a mindmap stand-in) when the incident spans people, process, tooling, and environment. Run Five Whys on the highest-signal ribs after brainstorming.