The 5 Whys is a root cause analysis method: state the problem, ask why it happened, then ask why of each answer until you reach a cause that, if removed, stops the problem from coming back. It usually takes about five whys, every answer must be a checked fact, and it should end at a process or system you can change, never at a person.
Free 5 Whys templateHow to do it5 worked examplesCheck the cause5 Whys vs fishbone
What Is the 5 Whys?
The 5 Whys came out of the Toyota Production System. Taiichi Ohno described asking why five times as the basis of Toyota's scientific approach: by repeating why, the nature of the problem and its solution become clear. The number is a habit, not a law. The point is to keep going past the first answer, which is almost always a symptom.
It needs no software and no statistics, which is its strength: a team leader can run one at the machine in fifteen minutes. It is also its weakness, because the answers are only as good as the facts behind them. The glossary entry has the one-line definition; the root cause analysis comparison shows where 5 Whys sits next to fishbone, fault tree, 8D, A3 and Kepner-Tregoe.
How to Do a 5 Whys Analysis
Write it on the free 5 Whys template, which keeps the problem statement, the chain, the root cause and the actions on one sheet and exports a PDF.
5 Whys Examples
Five illustrative examples from different settings. None is a client case; each shows the chain, the root cause, and how the action differs from the obvious quick fix.
Problem: The CNC on line 4 stopped three times this week on a spindle overload alarm while drilling part 7742.
- Why did the spindle overload? The 8 mm drill was dull and cutting forces rose.
- Why was the drill dull? It had drilled about 1,900 holes; the supplier rates it for 1,200.
- Why did it run past its life? No tool life limit was set for that tool in the controller.
- Why was no limit set? The program for 7742 was revised last month and released without tool life settings.
- Why was it released without them? The program release checklist does not include tool life.
Root cause: The program release process does not require tool life limits.
Action: Add tool life to the release checklist; set the controller to block cycle start when a tool has no life limit. Changing the drill was only containment.
Problem: Short shots on housing 220 rose to 3% of output, almost all in the first hour of Monday day shift.
- Why short shots? Not enough melt filled the cavity.
- Why not enough melt? Barrel zone 3 was 15 °C under setpoint when production started.
- Why was it under setpoint? Production started 10 minutes after power-on; after a weekend off the barrel needs about 25.
- Why start at 10 minutes? That is the warm-up the operators learned on weekdays, when the machine is already warm.
- Why is there no check? Start-up standard work says "warm up machine" with no temperature or time criterion.
Root cause: Start-up standard work has no measurable ready condition.
Action: Rewrite the start-up step as "all zones within 5 °C of setpoint" with a sign-off; add a controller interlock on barrel temperature.
Problem: Twelve orders missed the 4:00 pm carrier cut-off on Thursday and shipped a day late.
- Why did they miss the cut-off? Packing finished them at 4:40 pm.
- Why so late? Their pick wave was released at 3:30 pm.
- Why then? The planner releases the next wave only when the previous one is fully picked.
- Why that rule? The wave procedure was written when the site had one carrier with a 6:00 pm cut-off.
- Why was it not changed? Adding the second carrier did not trigger a review of wave planning.
Root cause: Wave release is not planned back from each carrier's cut-off.
Action: Schedule waves backwards from each cut-off (release by 2:30 pm for the 4:00 pm carrier); add procedure review to the carrier onboarding checklist.
Problem: A customer received size 10 fittings instead of size 12; the warehouse found 9 similar mispicks this month.
- Why the wrong size? The picker took the item from the slot next to the correct one.
- Why that slot? Sizes 10 and 12 sit side by side with nearly identical labels.
- Why side by side? Slotting groups each product family together for convenience.
- Why was the risk missed? The slotting rules do not consider look-alike items.
- Why was the pick not caught? Picks are confirmed by location scan only, not by item scan.
Root cause: Slotting ignores look-alikes, and pick confirmation does not check the item.
Action: Separate look-alike sizes by at least one slot; switch to scan-to-confirm on the item barcode.
Problem: Afternoon patients wait an average of 50 minutes past their appointment time.
- Why the wait? The doctor runs behind all session.
- Why behind? The session starts late: the first patient is not in a room at 1:00 pm.
- Why not roomed? One medical assistant does intake and also covers the phones at 1:00 pm.
- Why both? Phone volume peaks from 1:00 to 1:30 pm, the same time as first arrivals.
- Why was that not planned for? Staffing is set by head count, without looking at phone and arrival patterns.
Root cause: Staffing is not matched to the arrival and call pattern.
Action: Put a second person on phones from 1:00 to 1:30 pm; ask first patients to arrive 15 minutes early for intake.
In every example the first answer suggests a quick fix (change the drill, raise the shot size, retrain the picker) that would have worked for a week. The fifth answer points at the process that let it happen.
How to Check You Found the Real Root Cause
| Check | How | Example 1 (machine stop) |
|---|---|---|
| Facts, not guesses | Each answer is something you saw, measured or found in a record | Tool counter showed about 1,900 holes; supplier rating is 1,200 |
| Therefore test | Read the chain backwards: "X, therefore Y" | No life limit, therefore the drill ran 1,900 holes, therefore it went dull, therefore the spindle overloaded |
| Recurrence test | If you remove the last cause and change nothing else, does the problem stay gone for 90 days? | With the checklist and interlock, a new program cannot run without tool life |
| Detection leg | Also ask why the problem was not caught sooner | Why no warning before the overload? Spindle load monitoring alarm set at 100% instead of 85% |
| Not a person | If the chain ends at "operator" or "training", ask why the system allowed it | "Programmer forgot" became "the checklist did not ask" |
The detection leg is the idea behind the 3-legged 5 Whys: one chain for why it happened, one for why it was not detected, and one for why the system allowed it. 5 Why Pro builds all three legs as a branching tree with evidence and a verified tag on every answer, flags blame answers and chains that stop early, and exports into A3 Pro ($39, one-time, single user).
Limits of the 5 Whys
- One path. Problems where several causes combine (a fault tree's AND gate) get forced into a single chain. Branch when an answer has two true causes.
- Only as good as the room. If nobody present knows the process, the chain fills with plausible guesses. Bring the operator and go to the process.
- Different teams, different answers. Two teams can reach different root causes from the same problem. Data checks keep it honest.
- Drift into blame. "Why? Because Joe skipped the step" ends the analysis and starts a disciplinary process. Ask what made skipping the step possible or easy.
5 Whys vs Fishbone Diagram
| 5 Whys | Fishbone diagram | |
|---|---|---|
| Gives you | Depth: one chain to a root cause | Breadth: every plausible cause, by category |
| Use when | The likely cause is known or verified | Many possible causes, unclear where to look |
| Time | 15 to 30 minutes | 30 to 60 minutes with a team |
| Output | Root cause and actions | A list of suspects to check with data |
Use them together: a fishbone to list the suspects, data to pick the real one, then 5 Whys on it. The fishbone diagram guide walks through exactly that sequence on a leaking valve problem, and the fishbone diagram template is free to fill in. When the answer needs to go on one page for review, carry it into an A3.
5 Whys FAQ
What is the 5 Whys method?
The 5 Whys is a root cause analysis method: you state a problem, ask why it happened, then ask why again of each answer until you reach a cause that, if removed, stops the problem coming back. It came out of Toyota, where Taiichi Ohno taught it as a basic habit of problem solving. Five is a rule of thumb, not a rule.
What is a 5 Whys example?
A CNC machine stops on a spindle overload. Why? The drill was dull. Why? It ran past its life. Why? No tool life limit was set for that drill. Why? The new program was released without tool life settings. Why? The program release checklist does not include them. The fix is to add tool life to the release checklist and block cycle start when it is missing, not just to change the drill.
How do you know you have found the root cause?
Three checks. Read the chain backwards with "therefore" and make sure each step follows. Make sure every answer is a fact you checked, not a guess. And ask whether removing the last cause would stop the problem recurring; if the honest answer is "probably not", keep going or branch.
Do you always have to ask why five times?
No. Some problems reach an actionable cause in three whys, some need six or seven. Stop when you reach a cause you can act on that would prevent recurrence. Stopping at "operator error" or "lack of training" is almost always too early.
What are the limitations of the 5 Whys?
It follows one path, so problems with several causes acting together get oversimplified. It depends on what the people in the room know, and different teams can reach different answers. It drifts into blame if nobody insists on facts. For complex problems, start with a fishbone or fault tree and use 5 Whys on the verified causes.
What is the difference between 5 Whys and a fishbone diagram?
A fishbone diagram gives breadth: every plausible cause, grouped into categories such as the 6 Ms. 5 Whys gives depth: one cause followed down to its root. Use the fishbone when you do not know where to look, and 5 Whys once data has confirmed which cause to chase.
What is a 3-legged 5 Whys?
A version that asks three separate chains: why did the problem happen, why was it not detected before it reached the customer, and why did the system allow it. Each leg gets its own corrective action, which closes the gaps a single chain leaves open.