# The 5 whys: getting to the root cause of a problem

> The 5 whys is a chain of 'why?' questions that leads to a problem's root cause. Learn how to apply it, when to stop, and what its limits are.

*Tags: ux, technique, analysis, root-cause, product*

---


> [!info] Quick Definition
> The **5 whys** is a root cause analysis technique that asks "why?" in a chain, each time about the previous answer, until it reaches a cause that can be corrected. The number five is a guideline, not a rule.


## What is the technique?

Taiichi Ohno described it at Toyota as a way to reach the cause of a defect by repeating the question. Wikipedia adds that it was originally developed by Sakichi Toyoda and recounts a classic example: cross-threaded bolts in an engine block, whose chain of whys ends at a rusted shelf foot that destabilized the shelves and let parts fall, including the lost cutting tools. Mark Graban analyzes another Ohno example, a machine that stops because of a blown fuse and whose chain ends at a missing strainer. According to Graban, that chain really has seven whys, not five.

The core idea is simple: the symptom you see is rarely the cause. The trap is that the technique looks so simple that teams apply it without evidence and end up with a plausible answer that is not the right one.

For a designer it is useful because product problems usually show up as symptoms: "users do not finish sign-up", "support gets many calls about payment". Without asking why, it is easy to design a fix for the symptom, such as adding a help button, while the real cause is a confusing message or a validation that arrives too late. A why that points to a support or development process also shows that the problem is not only about the screen.

## How do you apply it?

A practical approach:

1. **Describe the problem** in one observable sentence, with data if you have it.
2. **Ask why it happened** and write the answer.
3. **Verify that answer** with evidence: a data point, a session clip, or the account of the person who lived it. It is not enough for it to sound right.
4. **Ask why again** about the verified answer.
5. **Stop when you reach a cause you can change**, not a circumstance you cannot control.

Illustrative example of a chain for a product problem: "The user does not pay" → "They cannot find the pay button" → "The button sits under an advertising block" → "The design of that screen was approved without a mobile review" → "There is no mobile review before screens are approved". The last cause is the one you can fix.

## When should you stop?

Wikipedia says five iterations are usually enough to reach a root cause that can be acted on, and that this cause is usually a process that is failing or missing; in the bolts case, for example, the absence of a regular shelf inspection.

Graban offers a practical test, which he attributes to Tracey Richardson: you have gone far enough when you can form a countermeasure that addresses the cause and all the symptoms up the chain. You have gone too far when you change the context of the problem entirely: her example about why she was late for work ends, absurdly, at "the Earth rotates". Graban adds, citing Jeffrey Liker, that you also go too far if you end up at something outside your control. The target is a controllable cause that will make an impact.

In design terms, that means stopping when the answer is no longer a decision you can make, such as "the law requires this data". At that point the chain has reached a useful limit.

## Is it a tool for finding someone to blame?

No, and it is worth saying so in the first meeting. Graban explains that, for Liker, the whys are a coaching process that should move people from blame to understanding the system. He sums it up as moving from "who screwed up?" to "what allowed this to happen?". If the last answer is a person, the chain has not ended.

TapRooT, a company that sells an alternative method, warns that the 5 whys can create an atmosphere of blame and suspicion because interviewees become defensive, and argues that blame gets in the way of analysis with any tool. That is a competitor's criticism, but it matches practice: ask about the process, not the person. "What step in the review let this through?" works better than "why didn't you check it?".

## What are its limits, and when is it not the right tool?

Wikipedia and TapRooT raise similar criticisms:

| Limit | What happens |
|---|---|
| One chain | It tends to isolate one cause, even though each question may have several possible answers |
| Stopping at symptoms | The team stops at the first plausible answer |
| Depends on the questioner | It does not take anyone beyond what they already know; without process knowledge, the chain drifts |
| Results are not repeatable | Two people can reach different causes in the same case |
| Arbitrary depth | According to Graban, in Ohno's case the number could have been three or six |

Alan Card, a medical professor cited in Wikipedia, criticized the method as a poor root cause analysis tool and suggested alternatives such as the Ishikawa (fishbone) diagram. Wikipedia also presents that diagram as one of the two ways to carry out the analysis, alongside a table format. If the problem has several causes, use the diagram first to see the full picture, then the 5 whys to dig into one branch.

It is not a good fit when the problem is complex and depends on many interacting factors, or when you have no evidence to verify the answers: a linear chain creates a false sense of certainty. It is also not useful for deciding what to build: the 5 whys analyze what happened, not what to create next.

## Example

Illustrative example: in an online store, returns for "wrong size" rose 20% after an update. A team applies the 5 whys. First, "the customer ordered a size different from the one they needed". Then, "the size guide does not match the product" and "the new brand's size chart changed without the team reviewing it". Finally, "the product upload process does not include a review of the size guide". That last cause can be fixed. The team adds a review step and, in the meantime, shows a warning on the new brand's product pages.

## Common mistakes

- **Applying it without evidence**, and accepting the first answer that sounds right.
- **Ending with a person** as the root cause.
- **Stopping at a symptom** because the answer seems sufficient.
- **Running the chain alone**, without the knowledge of the people who live the problem.
- **Not turning the cause into an action** with an owner and a date.
- **Using it for problems with many causes** without combining it with another tool.

To close the loop with actions and owners, the [post-launch review](/en/wiki/processes/post-launch-review/) offers a framework for turning findings into follow-up. And if the underlying question is how to change a problem more broadly, [How Might We questions](/en/wiki/techniques/how-might-we/) help move from cause to possible solutions.

## Sources

- [Five whys, Wikipedia](https://en.wikipedia.org/wiki/Five_whys)
- [Ohno's classic "5 Whys" example deserves another why, Lean Blog (Mark Graban)](https://www.leanblog.org/2026/04/ohno-5-whys-actually-seven/)
- [What's Wrong with 5-Whys?, TapRooT (vendor of an alternative method)](https://taproot.com/?p=98381)


---

Source: https://www.fernandoux.com/en/wiki/techniques/five-whys/
