Root Cause Analysis

Root cause analysis fails in a specific and predictable way: the investigation stops at the first plausible cause.

Something failed. Someone identifies a reason that could account for it. The reason is credible, the meeting has run long, there is pressure to close, and the investigation ends. The action follows from that cause, the record looks complete, and three months later the same failure returns — because the cause that was found explained the event but did not explain why the system allowed the event.

The default landing place for a stopped investigation is operator error. It is available, it is often technically true, and it ends the discussion. It is also almost never the root cause. If a competent person made a mistake, the real question is what about the tooling, the instructions, the layout, the workload or the checking permitted that mistake to occur and go undetected.

The discipline of root cause analysis is not intellectual sophistication. It is the willingness to keep going past the first satisfying answer.

• 5 Why — fast, verbal, effective on straightforward causal chains. Its weakness is that it produces a single line of enquiry, so a genuinely multi-causal failure gets forced into one narrative. Use it when the chain is likely linear.
• Ishikawa / fishbone — structures brainstorming across categories (people, method, machine, material, measurement, environment) so the team looks in places it would otherwise skip. Use it when the cause is genuinely unknown and you need breadth before depth.
• Pareto analysis — establishes which problem to investigate at all. Use it before the others, so effort goes where the loss is.
• 8D — the full structured discipline, with containment, root cause, corrective action, verification and prevention as distinct stages. Expected by automotive customers and appropriate for significant or customer-facing failures.
• Is / Is-Not analysis — comparing where the problem appears against where it does not, which is frequently the fastest route to a cause on intermittent failures.

Matching tool to problem matters. Teams that only know 5 Why apply it to complex multi-causal failures and produce confident, wrong answers. Teams that only know 8D apply heavy structure to trivial problems and exhaust their capacity on administration.

On your own failures, not on case studies. This is the whole approach. Textbook exercises have a designed answer, so teams learn to find the answer the exercise wants. Your real failures have ambiguous evidence, missing records, conflicting accounts and commercial pressure — and learning to reach a defensible cause under those conditions is the actual skill.

We facilitate live investigations with the people who were there. Root cause analysis conducted by one person at a desk reading records reliably misses what an operator would have said in the first two minutes. Part of what we train is how to run the conversation so people contribute rather than defend.

We also train the evidence discipline: distinguishing what is known from what is assumed, and holding a hypothesis as a hypothesis until it is tested. Most wrong conclusions come from an assumption that entered the discussion early and was never labelled as one.

Then we connect it to CAPA, because a cause identified and not acted on properly is wasted effort. We bring 30+ years and 900+ organizations to this, and it underpins the automotive 8D expectations in IATF 16949.

• Stopping at the first plausible cause
• Operator error as a conclusion rather than as a starting point
• Using 5 Why on a multi-causal failure, forcing it into a single narrative
• Investigating without the people who were present
• Assumptions treated as evidence because they entered the discussion early
• Skipping Pareto, so effort goes to the loudest problem rather than the largest
• Applying 8D to everything until the capacity to investigate anything properly is gone
• A cause identified but the action stopping at retraining
• No test of the hypothesis — the cause is asserted, never verified
• Investigations run as accountability exercises, which guarantees people withhold information

That final point undermines everything else. If people believe an investigation is looking for someone to blame, the information needed to find the cause will not be offered. How the investigation is framed determines what it can find.