Which records and relationships can a generated escalation dataset contain?

A generated escalation dataset can be built around a set of linked record types rather than isolated documents.

**Core records:** parties (individuals or businesses), applications or review items, existing facilities and balances, and the decisions requested.

**Policy records:** authority matrices by reviewer grade, product rules, exception categories, and the conditions under which a delegation applies. Policies can carry effective dates, so a case may be governed by a rule that has since changed.

**People and roles:** reviewers with grade, team, availability and declared conflicts, plus delegations that temporarily raise or move authority.

**Process records:** case notes, document checklists, escalation records with reason codes, returns for information, and audit events.

**Relationships that carry the difficulty:** a case links to a party, the party to several facilities, a facility to a guarantor, and a reviewer to the matrix row that governs them. Many escalation errors come from failing to follow one of these links.

A release can also include deliberate imperfections, such as a missing financial statement or a note that contradicts a field.

All of this is generated material informed by direct observation of how real businesses operate, not real business records. Buyers should inspect samples for grounding and disclosure risk, and scope the record types in a pilot.