What does a label that contradicts its own policy look like in QA?

A data operations analyst at a reseller is sampling a candidate release before listing it. They open `CLAIM-3391`, a contents claim after water damage.

The record shows an invoice, two photographs and a claim form. The assessment overall is `ready-to-proceed`, citing `RULE-0215`.

The analyst opens the rule. For contents claims above a stated threshold it requires proof of ownership, such as a purchase receipt or a bank record, in addition to the invoice. The claim value is above the threshold. No ownership document is in the case.

The label contradicts its own rule. The analyst records three findings:

- the assessment omitted a gap of code `missing` for proof of ownership - no request record exists, so the case would have moved forward without one - the rationale text mentions the invoice but never the threshold

The analyst checks nearby cases and finds the same pattern in a cluster that shares a rule version. The likely cause is a rule applied without its value condition.

The consequence if listed is that customers would train a model to approve claims that should be held, and would blame the model.

**What it teaches:** run re-derivation on a sample, group failures by rule version, and report the cluster back with identifiers. A fixed release then needs a re-check of that same cluster. A supplier that can show the rule and the case side by side makes this easy to do.