State the problem precisely before asking the first why
A vague problem statement produces vague root causes — be specific about what happened.
Why it works
The Five Whys chain is only as useful as its starting point. A problem stated as "things went wrong" or "quality was poor" can be explained by almost anything, so the first "why" branches in dozens of directions and the chain never converges. A precise problem statement — "the third weld in batch 47 failed at the seam, not the joint" — constrains the why-chain to the specific causal path relevant to this failure. Precision at the start is the investment that makes the rest of the analysis tractable.
How to do it
- Describe the problem as specifically as possible: what failed, where, when, under what conditions?
- Include what did NOT fail — this narrows the cause space.
- Avoid evaluative language ("it was bad") — stick to observable facts.
- If you are not sure of the precise facts, gather them before starting the why-chain.
Evidence
Problem specification is a foundational step in all formal root-cause analysis methods (FMEA, fault tree analysis, Ishikawa diagrams). Toyota’s own accounts of the Five Whys emphasize precise problem observation as the prerequisite. (mechanistic)
Precise specification is harder than it sounds in complex sociotechnical systems where the boundary of "the problem" is itself contested.
Sources
- Ohno (1988), Toyota Production System — problem specification as prerequisite to root cause analysis
Common mistake
Starting with a problem statement that is already a conclusion ("the process is inefficient") rather than an observation ("task X took 4 hours instead of 1"), which makes the first "why" circular.
Practice this with IX Coach
7 days free, then $40/month (~$1.30/day).
More practices for The Five Whys
- Follow the causal chain until you reach a changeable cause
Keep asking "why?" until you reach something you can actually fix.
- Branch the chain when you find multiple causes
When a "why" has two or more true answers, follow each branch separately.
- Distinguish human error from system conditions
When a person is the proximate cause, ask why the system allowed the error to occur.
- Verify the root cause by tracing back up the chain
After reaching a root, work back up: does each "because" in the chain make logical sense?
- Design the fix to prevent recurrence, not just to recover from this instance
Ensure the root-cause fix eliminates the cause permanently, not just this occurrence.