Receipt ledger
Every governed record should leave evidence
A receipt turns data-quality work from invisible cleanup into inspectable operational control. Refinery produces receipts for detection, decision, writeback, verification, and blocked actions.
What a receipt contains
- Record identity and governed path
- Decision outcome: accept, normalize, block, escalate, or review
- Policy version in force at decision time
- Issues detected: duplicate, invalid, missing, stale, conflicting, AI-unsafe
- Automatic actions applied under policy
- Blocked actions prevented from silent landing
- Human review requirement and approver when applicable
- Writeback status: pending, verified, failed, or blocked
- Evidence hash referencing supporting material
Why receipts matter
Prove why a duplicate merge was blocked or a field normalized before sync.
Show which records were marked unsafe before copilot or RAG consumption.
Export decision trails without exposing raw PII unnecessarily.
Example receipt
{
"record": "crm_account:acme-bv",
"path": "web_form_to_crm",
"decision": "normalized_and_escalated",
"policy_version": "customer-master-v3",
"issues": ["email casing", "possible duplicate", "missing industry"],
"automatic_actions": ["normalized email casing"],
"blocked_actions": ["duplicate merge", "industry overwrite"],
"human_review_required": true,
"writeback_status": "verified",
"evidence_hash": "sha256:..."
}
Privacy-aware evidence
Receipts can mask PII in customer-facing exports while preserving decision structure, policy version, and verification outcome for audit.
FAQ
Are receipts the same as data lineage?
Receipts are decision-level evidence for governed operational repair. Lineage tracks movement; receipts prove why a record was accepted, changed, or blocked.
Related pages
See baseline sample receiptsStart read-only. No production writeback required. Do not submit secrets through public website forms. Privacy