5 Whys Root Cause Analysis: Free Template, Examples & Steps (2026)
Quick answer: The 5 Whys is a root cause analysis technique where you start with a problem and ask "why did this happen?" again and again, usually about five times, until you reach a cause you can fix. Each answer becomes the next question. A good 5 Whys analysis starts with a clear problem statement, backs every answer with evidence, follows more than one branch when there's more than one cause, and stops at a process or system cause, never at "someone made a mistake". It ends with a corrective action that has an owner and a due date.
When something breaks, the first explanation is rarely the useful one. "The server ran out of disk" is true, but fixing it by deleting files only buys time. Ask why the disk filled up, why nobody noticed, and why there was no limit, and you reach something you can change for good.
That's the whole idea behind the 5 Whys. It needs no software, no statistics and no special training, which is why it's one of the most widely used root cause analysis methods, from factory floors to software postmortems.
This guide shows how to run a 5 Whys analysis step by step, gives you a free template, walks through real IT incident examples and the classic Jefferson Memorial story, explains how to keep it blameless in postmortems, and covers the mistakes that make the method fail.
What is the 5 Whys technique?
The 5 Whys is a simple method for finding the root cause of a problem. You state the problem, ask why it happened, then ask why that happened, and keep going until you reach a cause that, if fixed, would stop the problem coming back.
The number five is a rule of thumb, not a rule. Some problems reach a useful cause in three questions, others need seven. The point is to keep asking past the first, obvious answer.
Where it comes from
The method is credited to Sakichi Toyoda, the founder of what became the Toyota group, and was made famous by Taiichi Ohno as part of the Toyota Production System. Ohno's own example is still the clearest one:
Question | Answer |
|---|---|
Why did the machine stop? | A fuse blew because of an overload. |
Why was there an overload? | The bearing wasn't lubricated enough. |
Why wasn't it lubricated enough? | The lubrication pump wasn't pumping properly. |
Why wasn't the pump working properly? | Its shaft was worn and rattling. |
Why was the shaft worn? | There was no strainer, so metal scraps got in. |
Replacing the fuse would have restarted the machine for a day. Fitting a strainer stopped the failure for good. That difference, between fixing the symptom and fixing the cause, is what the 5 Whys is for.
When to use it, and when not to
Use the 5 Whys when | Use something else when |
|---|---|
The problem is specific and recent | The problem is vague, such as "quality is bad" |
One team can answer most of the questions | Many teams and systems are involved, with no shared view |
You need a quick, structured discussion | The causes interact in complex ways and need data to untangle |
You're running a postmortem on an incident | The stakes are high enough to need a formal method, such as in safety-critical industries |
For complex problems, the 5 Whys still works as a starting point, often alongside a fishbone diagram or a timeline. Our guide to root cause analysis compares the main methods.
How to do a 5 Whys root cause analysis
Bring the right people together. Include the people closest to the problem, such as the engineers who responded, plus a facilitator who keeps the discussion on track and doesn't push their own answer. Three to six people is usually enough.
Write a clear problem statement. Describe what happened, where, when and the impact, without guessing the cause. "Checkout failed for 18% of users between 14:02 and 14:41 UTC" is a problem statement. "The database was misconfigured" is already an answer.
Ask why it happened. Write down the answer the group agrees is true.
Check the answer against evidence. Logs, graphs, the incident timeline or a change record. If nobody can show it, mark it as a theory and verify it before going further.
Ask why again, and follow every branch. If an answer has two causes, follow both. Many problems have more than one root cause, and a single straight line will miss them. This branching version is often called a why tree.
Stop at a cause you can act on. A good stopping point is a process, design or system cause that your team can change, and that would stop the problem from recurring if fixed. A person's mistake is never a stopping point.
Ask "how" to fix each root cause: the 5 Hows. Turn each root cause into a corrective action: how do we prevent it, how do we detect it sooner, how do we limit the damage next time?
Give every action an owner and a due date, then check later that the problem really stopped recurring.
Running the session
Allow 30 to 60 minutes. Longer usually means the problem statement is too broad.
Write it where everyone can see it, on a whiteboard or a shared doc, so the chain is visible as it grows.
Ask "what" and "how", not "who". "What made this step easy to skip?" gets a better answer than "who skipped it?"
Read the chain backwards at the end. "No strainer, therefore metal in the pump, therefore a worn shaft…" If the logic breaks when read back, a step is missing.
In software teams, the 5 Whys usually happens inside a post-mortem meeting, after the incident is resolved and the timeline is complete.
5 Whys root cause analysis template
Copy the structure below into your incident doc or wiki.
1. Problem
Field | Your answer |
|---|---|
Problem statement (what, where, when, impact) | |
Date of the analysis | |
Facilitator | |
Participants | |
Links to evidence (incident record, timeline, graphs) |
2. The whys
# | Why did this happen? | Answer | Evidence | Verified? |
|---|---|---|---|---|
1 | Why did [the problem] happen? | ☐ | ||
2 | Why did [answer 1] happen? | ☐ | ||
3 | Why did [answer 2] happen? | ☐ | ||
4 | Why did [answer 3] happen? | ☐ | ||
5 | Why did [answer 4] happen? | ☐ |
If an answer has more than one cause, copy the table for each branch and label them A, B, C.
3. Root causes
Root cause | Branch | Within our control? | Would fixing it stop recurrence? |
|---|---|---|---|
☐ | ☐ |
4. Corrective actions (the 5 Hows)
Action | Prevents, detects or limits? | Owner | Due date | Status |
|---|---|---|---|---|
5. Follow-up check
Field | Your answer |
|---|---|
Date to review whether the problem recurred | |
Result |
The template works for any kind of problem. For software incidents, keep it with the postmortem and track the actions in your normal backlog; see postmortem action items.
5 Whys examples
Example 1: a checkout outage, with two branches
Problem statement: Checkout failed for 18% of users between 14:02 and 14:41 UTC on a weekday afternoon.
The first why splits in two, because there are two questions to answer: why did it fail, and why did it last 39 minutes?
Branch A: why did checkout fail?
# | Why? | Answer | Evidence |
|---|---|---|---|
1 | Why did checkout fail? | The orders database ran out of connections | Database dashboard: pool at 200 of 200 |
2 | Why did it run out of connections? | A long-running report query held connections open | Query log: report started 13:58 |
3 | Why was a report running on the orders database? | The new finance dashboard was connected to the primary database, not the read replica | Dashboard configuration |
4 | Why was it connected to the primary? | Nothing tells teams where read-only tools should connect, and the primary was the default | Onboarding docs |
Root cause A: there's no rule or safeguard keeping reporting workloads off the production database.
Branch B: why did it last 39 minutes?
# | Why? | Answer | Evidence |
|---|---|---|---|
1 | Why did it take 39 minutes to restore? | The on-call engineer was paged 9 minutes after the failures started | Alert and page timestamps |
2 | Why so late? | The alert watched database CPU, not checkout errors | Alert rule |
3 | Why did the alert watch CPU? | Alerts were set up per host years ago and never reviewed | Alert history |
Root cause B: alerting is based on server health rather than what users experience, and alerts are never reviewed.
Corrective actions (the 5 Hows):
Action | Type | Owner | Due |
|---|---|---|---|
Move all reporting tools to the read replica; give them a read-only user with a connection limit | Prevent | Data platform | 2 weeks |
Alert on checkout error rate instead of database CPU | Detect | Payments team | 1 week |
Review the noisiest and oldest alerts every quarter | Detect | SRE lead | Ongoing |
A single straight line would have found only one of these two causes. That's why it pays to branch.
Example 2: a missed page during maintenance
Problem statement: A critical alert went unanswered for 25 minutes during a planned database upgrade.
# | Why? | Answer |
|---|---|---|
1 | Why was the alert not answered? | The on-call engineer didn't notice it among 140 other pages that night |
2 | Why were there 140 pages? | Every expected alert from the upgrade fired normally |
3 | Why did expected alerts fire? | No maintenance window was set to hold them back |
4 | Why was no window set? | Setting one is a manual step that isn't in the change checklist |
Root cause: the change process doesn't include setting up alert suppression. Action: add a maintenance window step to the change template; see alert suppression and maintenance windows.
Example 3: the Jefferson Memorial
This is one of the most retold 5 Whys stories, used in training for decades. Details vary between tellings, but the usual version goes like this:
# | Why? | Answer |
|---|---|---|
1 | Why is the memorial's stone wearing away? | It's washed very often with harsh cleaners. |
2 | Why is it washed so often? | To remove bird droppings. |
3 | Why are there so many birds? | They come to eat the spiders. |
4 | Why are there so many spiders? | They come to eat tiny insects. |
5 | Why are there so many insects? | They're drawn to the lights, which come on at dusk. |
The fix in the story is to switch the lights on a little later in the evening. The lesson is the same as in the software examples: the first answer, "use gentler cleaners", treats the symptom, while the fifth answer removes the cause.
Using the 5 Whys in blameless postmortems
In software teams, the 5 Whys is most often used after an incident. That's also where it most often goes wrong, because the easiest answer to "why?" is a person.
Rule 1: "Human error" is never a root cause. If an answer is "the engineer ran the wrong command", ask one more question: why was it possible to run that command, against that system, without a check? People make mistakes in every system. The question is why the system let this one cause an outage.
Rule 2: ask about conditions, not people.
Instead of | Ask |
|---|---|
Who deployed the bad change? | What let a bad change reach production? |
Why didn't the on-call engineer notice? | What made the signal hard to notice? |
Why did Sam skip the step? | What made the step easy to skip? |
Rule 3: follow the detection and response branch too. Most incidents have at least two questions: why did it happen, and why did it last as long as it did? The second branch often leads to cheaper fixes, such as better alerts, clearer runbooks or faster rollback.
Rule 4: use the timeline as evidence. Every answer should point to something: a log line, a graph, a change record, a timestamp. A complete incident timeline makes this fast.
Rule 5: end with actions, not findings. A postmortem that names root causes but assigns no actions changes nothing. Each root cause needs at least one corrective action with an owner and a due date.
Where a root cause is bigger than one incident, for example the same weakness behind several outages, hand it to problem management. In ITIL terms, that's where the 5 Whys becomes part of problem management rather than incident management.
Common 5 Whys mistakes
Mistake | What it looks like | How to avoid it |
|---|---|---|
Stopping too early | "Why did the site go down? The server crashed." End. | Keep asking until you reach something you can change that would stop it recurring |
Stopping at a person | "Because the engineer made a mistake" | Ask why the system allowed the mistake to cause harm |
Guessing instead of checking | Answers based on what people think happened | Back every answer with evidence; mark guesses as theories |
Following only one branch | One neat chain for a problem with several causes | When an answer has more than one cause, follow each |
A vague problem statement | "Performance is bad" | State what happened, where, when and the impact |
Forcing exactly five | Inventing a fifth answer, or stopping at five when more is needed | Five is a guide; stop when the cause is actionable |
Going too far | "Because the company didn't hire enough people in 2019" | Stop at a cause within your team's control |
No follow-through | Root causes listed, nothing changes | Assign owners and due dates; check later that the problem stopped |
Limitations of the 5 Whys
The method's simplicity is also its weakness. It depends on what the people in the room know, so it can miss causes nobody present can see. Different groups can reach different answers for the same problem. And it handles problems with many interacting causes poorly, because each chain looks at one thread at a time.
None of this makes it less useful. It means the 5 Whys works best for specific problems, with evidence at each step, with branches when the causes split, and alongside other methods when the problem is genuinely complex.
5 Whys vs fishbone and other root cause analysis methods
Method | How it works | Best for | Effort |
|---|---|---|---|
5 Whys | Ask "why" repeatedly, following each branch | Specific problems with a traceable chain of causes | Low: one meeting |
Fishbone (Ishikawa) diagram | Brainstorm possible causes under categories such as people, process, tools and environment | Problems where you don't yet know where to look | Medium |
Fault tree analysis | Work backwards from a failure through a logic tree of AND and OR conditions | Safety-critical or complex systems | High |
Change analysis | Compare what was different between when it worked and when it failed | Failures that appeared after a change | Low |
Timeline analysis | Reconstruct the sequence of events with timestamps | Any incident; the evidence base for every other method | Medium |
5 Whys vs fishbone in one sentence: a fishbone diagram goes wide to list possible causes, the 5 Whys goes deep to follow one of them to its root. They work well together: use the fishbone to find the candidate causes, then the 5 Whys to drill into the ones the evidence supports.
For how these methods fit into a complete root cause analysis, see our root cause analysis guide.
Where ITOC360 fits
A 5 Whys analysis is only as good as its evidence. ITOC360 Incident Response collects that evidence while the incident is happening:
A timestamped timeline of alerts, pages, acknowledgments, escalations and decisions, so every "why" can point to what actually happened.
Correlated alerts that show which signals fired first and which followed, useful for separating causes from symptoms.
Detection and response times for every incident, which make the "why did it last so long?" branch easy to answer.
The analysis itself stays with your team, in the postmortem. ITOC360 makes sure the facts are there when you start asking why.
Frequently asked questions
What are the 5 Whys in root cause analysis?
The 5 Whys is a technique where you ask "why did this happen?" repeatedly, usually about five times, starting from a problem and moving through each answer until you reach a root cause you can fix.
How do you do a 5 Whys analysis?
Write a clear problem statement, ask why it happened, check the answer against evidence, then ask why again. Follow every branch when an answer has more than one cause, stop at a cause your team can change, and turn each root cause into a corrective action with an owner and a due date.
Do you always have to ask why five times?
No. Five is a rule of thumb. Some problems reach a useful root cause in three questions, others need seven or more. Stop when you reach a cause that, if fixed, would stop the problem recurring.
What is a 5 Whys example?
The classic example is Taiichi Ohno's: a machine stopped because a fuse blew, because of an overload, because a bearing wasn't lubricated, because a pump was failing, because there was no strainer to keep metal scraps out. Fitting a strainer fixed the problem for good.
Who invented the 5 Whys?
The technique is credited to Sakichi Toyoda, founder of what became the Toyota group, and was made widely known by Taiichi Ohno through the Toyota Production System.
What are the 5 Whys and 5 Hows?
The 5 Whys finds the root cause. The 5 Hows turns it into a solution by asking how to prevent the problem, how to detect it sooner and how to limit the damage next time.
What is the difference between the 5 Whys and a fishbone diagram?
A fishbone diagram lists many possible causes under categories. The 5 Whys follows one cause deep to its root. Teams often use a fishbone to find candidate causes and the 5 Whys to drill into the ones the evidence supports.
What are the limitations of the 5 Whys?
It depends on what the participants know, different groups can reach different answers, and it handles problems with many interacting causes poorly. Backing each answer with evidence and following every branch reduces these weaknesses.
Is there a free 5 Whys template?
Yes. This page includes a copyable 5 Whys root cause analysis template, with Word, Excel and PDF downloads.
Sources
Taiichi Ohno, Toyota Production System: Beyond Large-Scale Production (the machine and strainer example)
Institute for Healthcare Improvement: 5 Whys: Finding the Root Cause