Five Whys, All the Way Down
Get from the symptom to the mechanism instead of stopping at 'careless'
Trace this problem with five whys: [describe the symptom]. Rules: 1) one most likely cause per level, written as a single 'because X, therefore Y' sentence; 2) after each level, label the evidence strength (data / inference only); 3) when evidence is missing, don't push the chain further — give me a check I can run within a week; 4) stop at level five, output the full causal chain, and mark which level is actually actionable.
How to use
Around the third why it slides into 'people weren't careful'. Making it label evidence strength blocks that slide.
Good for
- The same problem shows up for the third time this month — an outage, a late delivery, rework — and every fix ends with 'we'll be more careful'
- You already have a list of remedies (more training, more ownership) and nobody can say which link in the chain each one touches
Not for
- The problem just needs containing first — a single bad invoice, one broken machine — and root-cause work can wait
- There's no history to learn from, so asking why again turns into speculation about someone's motives
Example
Input
Payment success rate dropped from 98% to 91% after a release. A restart fixed it. Third time this quarter.
What you get
It should walk from callback timeouts to a backlog of retry jobs to the queue hitting its ceiling to scaling that added machines but not queue capacity, and stop at the capacity review that never counted retries. Each level should be labeled as measured or inferred. If the third level has no logs, the right output is a logging check to run this week, not another invented cause.
Common mistakes
- Turning why into an accusation ('why didn't you think of this'), which yields defensiveness and excuses instead of mechanisms
- Filling all five levels with inference so the chain looks complete, and the fix lands on whatever is easiest to change
Try tweaking it like this
- Rephrase as 'what conditions all had to hold for this to happen', which fits multi-factor incidents better than a single chain
- Add 'counterfactual: if we had done X, would this still have happened?', which separates root causes from background conditions