Case: Functional Outlier
One Part Failed, Seven Did Not
This review asks whether one failing part is random contamination or a first warning of repeatability loss in the run.
Open Case StudyOutlier Decision Matrix
A single anomaly does not automatically reject a batch. We score outliers by repeatability risk, functional impact, and evidence strength.
| Signal | Typical Reading | Action |
|---|---|---|
| Isolated visual defect | 1 of 8 parts | Document exception, release with note |
| Fit or load failure | Any single part | Hold batch, expand sample |
| Pattern recurrence | 2+ matching anomalies | Treat as process drift |
Case: Late-Batch Signal
A Later Sample Changed the Final Decision
An outlier appeared only in later samples, changing acceptance logic and proving that timing matters in small-batch evidence.
Open Case Study