Nonconformity Root-Cause Assistant

What it does

It rescues a nonconformity from the shallow "operator was careless → training given" closure and turns it into a CAPA draft that survives an audit. The flow is anchored to canon, not intuition:

1. 5W2H problem objectification — before any RCA, the finding is made measurable via What/Where/When/Who/Which/How/How-many ("sometimes drifts" → "2 of last 3 lots exceeded +0.4 mm tolerance"), preventing root-cause hunting on the wrong problem.
2. Ishikawa / Fishbone + 6M scan — candidate causes are distributed across Man, Machine, Method, Material, Measurement, Environment; each hypothesis is either eliminated or accepted with evidence (auditors also want the "why not" rationale).
3. occurrence / detection / systemic triad — most nonconformities have two chains: why it occurred and why our control failed to catch it. The 5-Why tree runs these branches separately, validating each link with a reverse "therefore-test"; an unevidenced link breaks the chain and the assumption is written to gaps.
4. FMEA-derived risk scoring — each candidate root gets RPN = Occurrence × Detection × Severity (1-10); RPN ≥ 100 or S ≥ 9 triggers management/HITL escalation.
5. Error-proofing hierarchy CAPA — eliminate > poka-yoke/mistake-proof > automatic control > procedure+checklist > training+alert, in that order; training is the weakest, last resort. Each root gets correction (immediate) + corrective (permanent) + detection action + a measurable effectiveness verification (numeric criterion + date + responsible role).

The result: a 0-100 RCA Depth/Quality score, a Shallow/Adequate/Audit-defensible verdict, ranked findings and a CAPA table bound to the standard by clause number.

When to use it

When there is an internal/external audit finding, customer complaint, process deviation or lab result breach and you need a standard-compliant CAPA record that prevents recurrence. Especially when you want to avoid the most common major-finding causes — mistaking a correction for a root cause, closing without effectiveness verification, and the human-blame wall. Built for quality managers, accreditation officers, lab quality representatives and internal auditors.

Method / frameworks

5 Whys (Toyota / Ohno), Ishikawa Cause-and-Effect + 6M (in lab context 5M1E + ISO/IEC 17025 measurement uncertainty), Apollo RCA / Cause-Mapping multi-cause logic (every effect has ≥2 causes), Pareto 80/20 prioritization, FMEA RPN (O×D×S) risk scoring, and the 8D/PDCA loop. Standard anchors: ISO 9001:2015 §10.2, ISO/IEC 17025:2017 §7.10/§8.7/§6.4.10, ISO 13485:2016 §8.5 / FDA 21 CFR 820.100, IATF 16949 + AIAG-VDA. Clause numbers are verified against the edition in force; an unverifiable clause is never fabricated — it goes to gaps.