Key takeaways
- RCA finds the underlying cause so a problem stops coming back, instead of patching symptoms.
- 5 Whys for simple, single-cause problems; Fishbone when several causes may combine.
- Stop at a cause you can actually control and verify, not at "operator error".
- Good RCA needs good evidence: what happened, when, and how often.
Most repeat breakdowns are not new problems, they are the same problem returning because the last fix treated a symptom. Root cause analysis is how you break that loop: dig past what failed to why it failed, then remove that.
The 5 Whys
State the problem, then ask "why" of each answer until you reach an actionable cause. A worked example:
| Step | Answer |
|---|---|
| Problem | The conveyor stopped and the line went down. |
| Why? | The drive motor tripped on overload. |
| Why? | The belt jammed against a build-up of material. |
| Why? | Spillage from the transfer point accumulated under the belt. |
| Why? | The skirting and belt cleaner were worn and not replaced. |
| Why? (root) | There was no inspection task for skirting/cleaner wear. |
The fix is not "restart the conveyor" or "tell the operator to watch it". It is adding the inspection task. That removes the cause. (See the conveyor troubleshooting guide for the symptom-to-fix detail.)
Limit: the 5 Whys assumes one linear chain. If a problem has several contributing causes, it can lock onto one and miss the rest. That is when you reach for a Fishbone.
The Fishbone (Ishikawa) diagram
Draw the problem as the fish head, then add "bones" for each category of possible cause and brainstorm under each. The classic six categories (the "6 Ms"):
| Category | Ask |
|---|---|
| Machine | Equipment, tooling, wear, settings |
| Method | Process, procedure, changeover, sequence |
| Material | Incoming quality, variation, handling |
| Manpower | Training, staffing, fatigue, instructions |
| Measurement | Gauges, calibration, what we record |
| Environment | Temperature, humidity, dust, vibration |
You do not fix everything on the diagram. You use it to find the few likely causes, then test them with evidence and fix the ones that prove out.
Describe the symptom to the AI troubleshoot assistant or search 100+ documented problems.
The mistakes that ruin RCA
- Stopping at "human error". Almost always there is a system reason the error was possible. Keep going.
- Guessing without evidence. RCA needs facts: what happened, when, how often. If stops are not captured accurately, the analysis is built on sand.
- One and done. Verify the fix actually reduced recurrence. Track it.
- Blame over system. The goal is a cause you can control, not someone to fault.
The hard part: getting the true cause
RCA is only as good as the evidence behind it, and on a fast line the evidence is exactly what goes missing. Short micro-stops are cleared in seconds and never logged, so the team ends up guessing. That is the gap the partner we recommend, , closes: it captures every stop from the PLC and shows the true cause on video automatically, then routes a work order. It is EU-built with EU data residency and holds ISO 27001 / 20000-1 / 9001 (supports audit-readiness). Fabrico is a partner we recommend; this guide and the tools are free regardless.
How many "whys" do I really need?
Five is a guideline, not a rule. Stop when you reach a cause you can control and verify, whether that is three whys or seven.
What about Pareto and FMEA?
Pareto helps you pick which problem to analyse (the vital few causing most of the loss). FMEA is proactive: it ranks potential failure modes before they happen. Both pair well with 5 Whys and Fishbone.
Who should be in an RCA?
The people closest to the work, operators and maintenance, plus anyone who owns the process. Diversity of view is what makes the Fishbone work.