Root cause analysis

The move from complaint to cause#

A complaint is an observation filtered through one person's vocabulary. Two customers can describe the same underlying failure in language that shares no words. Poth treats the complaint as evidence about a cause, not as the cause itself.

What gets compared#

• Feedback signal — what customers reported, in their own words
• Product data — what they actually did in the product, and when
• Operational data — what your team logged as done on the other side
• Behavioural patterns — how the affected segment differs from everyone else

A complaint that survives contact with operational data is a real problem. One that does not is usually a symptom of something else, and that something else is what you want.

A worked example#

A facility receives repeated complaints about cleanliness. The obvious explanations are general standards and missed cleaning rounds. Operational logs show morning cleaning completed on every relevant day, so both are rejected at low confidence.

What survives: supplies run to near zero by Thursday, and the supplier delivers on Monday only. The cause is a cadence mismatch, not a cleaning failure. The fix is a delivery schedule, and no amount of re-training staff would have found it.

Why this changes the action#

Acting on the complaint would have produced a cleaning audit. Acting on the cause produces a supply schedule change. Same feedback, different bill, different outcome.

Next#

• Hypothesis engine explains how competing explanations get scored.
• Act on an insight covers assigning the resulting decision.