Skip to content
MirrorFirm

Which record types should I ask a generated allocation dataset to include?

A checklist of the record types, relationships and supporting artefacts to name in a request, plus a note on asking how each was generated.

When you request generated material for this workflow, name the record types and relationships up front; this list covers the usual ones.

- Open items with status, due date and open amount, so partial settlement can be represented. - Bank lines with payer text, reference text and value date, including truncated and noisy references. - Remittance advice in both structured and free-text form. - Customer and related-entity links, including groups that pay on behalf of members. - Credit notes, refunds and reversals with approval state. - Written policies covering tolerances, suspense holding and escalation owners. - Tool definitions for search, post, park and escalate actions. - Reference outcomes per task, with the rule each one relies on.

Ask the provider to say which of these are generated from scratch and which are derived from other generated records. That distinction affects how you should test for internal consistency. As the output is synthetic, also ask how the generation was grounded and what checks were made for disclosure risk, then run your own spot checks on the sample before agreeing to anything larger.

Naming record types and relationships up front prevents gaps that only show up once your harness tries to score a run.

Audience:
Evaluation teams
Content type:
Checklist
Buyer question:
What can be generated

Content release: r-a220eb6e099b243b

More on Payment allocation