Classify each return by the defect, its likely origin and the evidence supporting that attribution. A return reason such as “incorrect scope” is not enough to establish whether the inspector, producer, rule owner or integration caused the problem.
Use this proposed rework taxonomy:
| Example return | Possible origin | Evidence needed before attribution | Potential corrective action |
|---|---|---|---|
| Missing measurement | Inspection | Original submission and intake requirements | Improve capture instructions or intake screening |
| Quantity differs from supplied sketch | Production | Source dimension, calculation and estimate history | Correct calculation or mapping logic |
| Carrier requirement applied inconsistently | Rule governance or production | Effective guideline version and assignment date | Clarify the rule or improve training |
| Photograph missing after transfer | Integration | Source inventory and transfer logs | Repair the attachment hand-off |
| Estimate changes after newly discovered damage | New information | Evidence-arrival and inspection chronology | Treat as a supplement where appropriate |
| Conflicting technical opinions | Unknown or judgement | Independent technical review | Adjudicate before assigning blame |
Maintain these fields: pseudonymous file ID, inspection source, production route, complexity, defect code, severity, detection stage, suspected origin, confirmed origin, correction minutes and resolution.
Rank improvement opportunities by confirmed frequency × average attributable cost, while separately tracking consumer-impacting defects. If denominators or comparable case mix are unavailable, do not publish a supplier league table.
“Unknown” is a valid result. False certainty about causation can send procurement towards the wrong replacement supplier.