White Belt · 11 min
5 Whys: knowing when you have reached a cause you can act on
After this you can
- run a 5 Whys chain and recognise when it has reached a cause I can act on.
Assumes you have done Pareto: finding the few categories that carry the problem.
The problem
A team traced a recurring order error through five whys and arrived at "because the operator was not paying attention". They ran a retraining session. Four months later the same error was still happening at the same rate — and the operator had left. The chain had five links and stopped at the first one that could be blamed on a person, which is the most common way this tool fails.
The idea
You ask why? repeatedly, tracing a problem from its visible symptom back to a cause you can actually change. Taiichi Ohno described it as the basis of Toyota's scientific approach: "by repeating why five times, the nature of the problem as well as its solution becomes clear."
Five is a rule of thumb, not a target. Some chains end at three. Some need seven. Stopping because you reached five is as wrong as stopping because you reached a person.
Two tests tell you when to stop.
The reverse read. Walk the chain backwards saying "therefore" instead of "why". "There is no filter on the pump intake, therefore grounds enter the housing, therefore the bearing wears, therefore the pump draws too much current, therefore the cut-out trips, therefore the machine stops." If any link makes you hesitate, that link is wrong — you have jumped, or guessed. This single check catches most bad chains, and almost nobody does it.
The actionability test. Can you change this thing, and would changing it stop the effect? "The operator was not paying attention" fails both: you cannot reliably change attention, and a different operator will produce the same error tomorrow. That answer is not a root cause, it is where the chain gave up.
Three failure modes to watch for:
- Stopping at a person. Whenever a why lands on someone's behaviour, ask one more: why was it possible for that to cause a problem? A process that a moment's inattention can break is a fragile process, and that is the finding.
- A single chain for a multi-cause problem. 5 Whys traces one thread. If your problem has several contributing causes, you need a fishbone or a proper data analysis. A single chain will confidently produce one answer and hide the rest.
- Confusing plausible with verified. The chain is a hypothesis. Each link should be checkable, and the important ones should actually be checked. A chain that has never met data is a story the team told itself.
Worked example
The office coffee machine stops mid-brew three or four times a week.
Why does it stop? The overload cut-out trips. Why does the cut-out trip? The pump draws more current than its rating. Why does the pump draw too much current? Its bearing is worn. Why is the bearing worn? Grounds are getting into the pump housing. Why are grounds getting in? There is no filter on the pump intake.
Five links. Now do the work most teams skip — read it backwards with "therefore":
No filter on the intake → therefore grounds enter the housing → therefore the bearing wears → therefore the pump draws too much current → therefore the cut-out trips → therefore the machine stops.
Every link holds. No hesitation anywhere. The chain is sound.
Actionability. Can you fit a filter? Yes, for a few pounds. Would it stop the effect? Yes — no grounds, no wear, no over-current, no trip.
Compare that with where the chain would have ended two links earlier, at the third why. "The bearing is worn" → replace the bearing. Perfectly reasonable, completely useless: without the filter, the new bearing wears out on exactly the same schedule and you have bought yourself a maintenance subscription rather than a fix.
That is the difference between a cause and a root cause, and here it is two whys apart.
Now notice what the chain never mentions: nobody. No operator, no cleaner, no maintenance technician. A good chain usually ends at a design decision, and this one ends at a filter that was never fitted.
Dataset: ds-coffee-machine — the same data loads in the tool below, so you can reproduce every figure here yourself.
Your turn
The chain starts empty with only the problem filled in. Build the coffee-machine chain yourself, link by link — the tool asks you to confirm the reverse read at each one before it lets you add the next, which is the backwards test applied mechanically rather than remembered.
Then try this: stop after the third why, at "the bearing is worn", and write down the action you would take. Now build the remaining two and write down that action. The tool does not compare them for you; the comparison is the exercise. One buys a bearing every few months and the other fits a filter once.
Then start a new chain for a problem in your own work. When you hit a why that names a person, do not stop — ask why the process allowed that to matter.
5 Whys Chain
practiceWhen to stop
Not at five. Five is a rule of thumb — some chains end at three, some need seven. Stop when the cause is something you can change, and changing it would stop the effect.
A good chain usually ends at a design decision, not at a person. If yours ends at someone’s behaviour, it stopped one why early.
Source: Ohno (1988), Toyota Production System — “by repeating why five times, the nature of the problem as well as its solution becomes clear.”
Check yourself
No hints. Wrong answers are explained, not softened.
A chain ends at: 'because the technician forgot to log the reading'. What should you do?
Reading a chain backwards with 'therefore', you hesitate between link 2 and link 3. What does that mean?
A team's Pareto shows three roughly equal defect categories with different mechanisms. They start a 5 Whys on the largest. What is the risk?
Worth remembering
The reverse read
Walk the chain backwards saying 'therefore'. If any link makes you hesitate, that link is a jump or a guess.
When a why lands on a person
Do not stop. Ask why the process was fragile enough for that behaviour to cause a problem.
When is 5 Whys the wrong tool?
When a problem has several contributing causes. One chain yields one answer and hides the rest — use a fishbone or data analysis.
Can you do this now?
Rate yourself honestly. We compare your rating with how you actually answered — the gap is more useful than either number alone.
I can run a 5 Whys chain and recognise when it has reached a cause I can act on.
Your rating is recorded alongside your drill results. Neither alone marks the competency as met.