Practitioner guide & template
5-Whys root cause analysis — a practitioner's guide (plus a free template)
5-Whys is the fastest way to move a safety investigation from symptom to systemic cause — when it is done properly. This guide covers the method Toyota actually used, the failure modes that make most 5-Whys charts useless in an audit, a fully-worked example with evidence trail, and how AI can widen your causal frame without replacing the investigator's judgment.
Last reviewed 9 July 2026 · SafeGuard EHS Editorial · 9 min read · Companion to our incident tracking guide.
Method
Five steps that turn a symptom into a systemic root cause
Each step exists to prevent a specific failure mode we see in real audit sampling.
Anchor the problem statement in observable facts
Before the first 'why', write a one-sentence problem statement using only what was observed. Include what, where, when, and how much. Skip who and why. 'Operator sustained a laceration to the right hand at 14:20 on line 3 during blade changeover' is anchored; 'blade changeover injuries are increasing' is not.
Ask 'why' about the previous answer, not the original problem
Each 'why' must interrogate the answer you just wrote — not restate the original event. This is the single most common failure mode: teams write five parallel reasons for the same symptom instead of a causal chain. If your Why 3 could logically come before Why 2, the chain is broken.
Stop at a systemic cause, not a person
'Operator did not follow the SOP' is where weak 5-Whys stop. The next 'why' is the point: was the SOP present, current, trained, and enforceable in the actual work conditions? Root causes are always about the system that let a mistake happen — training, design, supervision, procedure, culture — never about individual blame.
Validate each link with evidence
Every 'because' needs a source — a maintenance log, training record, CCTV timestamp, interview quote, or measurement. A chain built from opinion collapses under audit. If a link cannot be evidenced, mark it as a hypothesis and add a verification action.
Convert the root cause into corrective and preventive actions
The output of a 5-Whys is not the chart — it is the CAPA. Each identified systemic cause needs at least one corrective action (fix this instance) and one preventive action (stop recurrence). Assign an owner, a due date, and a verification method. Anything else is a paper exercise.
Worked example
Blade-changeover laceration — the full chain with evidence
A real-shaped example showing why the root cause sits at the Management of Change procedure, not the operator.
| Step | Answer | Evidence |
|---|---|---|
| Problem | Operator sustained laceration to the right hand at 14:20 on line 3 during blade changeover. | Incident report #INC-2026-118, first aid log |
| Why 1 | Because the blade guard interlock was bypassed during the changeover. | Photo of bypassed interlock, maintenance log |
| Why 2 | Because the changeover SOP requires guard removal but does not require re-engaging the interlock until the machine is closed. | SOP-LN3-CHG rev 4, section 3.2 |
| Why 3 | Because when the SOP was written in 2021, blade changeovers were performed with the machine fully de-energised — the interlock question did not arise. | SOP change log, engineering change record ENG-19-072 |
| Why 4 | Because the 2023 upgrade to a hot-swap blade cartridge changed the physical procedure but the SOP was not updated to reflect the new interlock exposure. | ENG-23-041 project file, MoC record MOC-23-118 |
| Why 5 — Root cause | Because Management of Change (MoC) does not require an SOP + risk-assessment review for equipment upgrades classified as 'like-for-like replacement'. | MoC procedure MOC-P-01 rev 6, section 4.1 |
| Corrective action | Update SOP-LN3-CHG to reflect hot-swap procedure and interlock re-engagement. Owner: Ops Manager. Due: 14 days. Verify: SOP review + operator sign-off. | CAPA-2026-441 |
| Preventive action | Amend MoC-P-01 to require SOP + risk-assessment review for any equipment upgrade that changes operator interaction, regardless of like-for-like classification. Owner: HSE Director. Due: 60 days. Verify: audit of next 5 MoCs against revised criterion. | CAPA-2026-442 |
Notice where a weak chain would have stopped: Why 1 (guard bypassed) or Why 2 (SOP gap) both feel like conclusions. The systemic root cause is at Why 5 — a procedural classification in Management of Change that let equipment behaviour change without an SOP + risk-assessment review.
AI-assisted RCA
Where AI genuinely helps — and where it doesn't
Use large language models to widen the causal frame and stress-test evidence. Do not use them to write the finding.
Four prompts that improve a 5-Whys session
Prompt 1 — Widen the causal frame
Here is my 5-Whys chain for [incident]. Suggest three alternative branches for Why 2 I may have missed — consider procedure, equipment, training, supervision, and environment. Flag any branch that would need different evidence to verify.
Prompt 2 — Stress-test each link
For each 'because' link in the chain below, list the evidence type that would prove or disprove it (record type, measurement, interview question). Rate each link's evidence strength as strong / weak / unverified.
Prompt 3 — Rewrite blame-shaped answers as system answers
Rewrite any answer in this chain that assigns cause to an individual as a system-level cause instead (procedure gap, training gap, design gap, supervision gap, cultural gap). Explain what changed.
Prompt 4 — Draft CAPAs from the root cause
Given the root cause '[X]', draft one corrective action and two preventive actions. Each must have a measurable success criterion, a suggested owner role, and a verification method (audit, sampling, KPI).
What AI must not do
- Assign the finding — accountability rests with a named investigator.
- Replace evidence — every link still needs a verifiable source.
- Interview witnesses — human judgment on tone, hesitation, and site context does not transfer.
- Sign off CAPAs — verification is a human act, tied to a role with authority to close the incident.
Common mistakes
Five failure modes that make 5-Whys charts useless
Stopping at 'operator error'
Human error is almost never a root cause — it is the point where the investigation should get interesting. The next Why is: what let the mistake happen (procedure, design, training, supervision, culture)?
Writing parallel symptoms as a chain
Every 'Why' must interrogate the previous answer, not the original event. If Why 3 could logically come before Why 2 without changing meaning, the chain is broken.
Skipping evidence
A chain built from opinion collapses under audit. Every 'because' needs a record type — log, photo, interview, measurement — or it is a hypothesis, not a finding.
Confusing corrective action with root cause
'We retrained the operator' is the fix, not the finding. The root cause is what let the training gap exist — usually a competence management or supervision system issue.
Treating the chart as the deliverable
The deliverable is a CAPA with owner, due date, and verification. A 5-Whys that does not produce an assigned, verifiable action is a filing exercise.
Running it alone at a desk
The people involved know things the report does not capture. A facilitator who did not witness the event is essential to challenge assumptions. Six people, within 48 hours.
Free template
Download the blank 5-Whys template
A one-page CSV you can open in Excel, Google Sheets, or paste into an incident record.
5-whys-template.csv
Problem, five Whys, corrective + preventive action, verification, owner, due date.
5-whys-worked-example.csv
The blade-changeover chain from this guide, evidence trail included.
Related topics
Where 5-Whys sits in your program
Incident tracking
How 5-Whys attaches to the incident record and drives corrective actions to closure.
Safety audits
What auditors look for when they sample your root cause analyses.
AI-assisted RCA in SafeGuard EHS
See how the platform links 5-Whys chains to incidents, CAPAs, and effectiveness verification.
FAQ
5-Whys root cause analysis — common questions
Run 5-Whys inside your incident record — not in a spreadsheet
SafeGuard EHS attaches a structured 5-Whys chart to every recordable incident, uses AI to widen the causal frame, and links each root cause to a tracked corrective action with an owner, due date, and verification method.
Start here