IfThenWhy™ Framework Audit Check Sheet
Asset ID: ____________________ | Auditor: ____________________ | Date: __________
Pillar 1: Validation (The Audit)
[ ] Result Verification: Have the automated tests (Validation_1_Logic) returned a True result in the manifest?
[ ] Anomaly Stress Test: Does the logic pass a "zero-record" filter without crashing?
Pillar 2: Persona (The Stakeholder)
[ ] Stakeholder Attribution: Is the metric anchored to a specific owner? (e.g., John.Doe@google.com).
[ ] Intent Alignment: Does the implementation satisfy the "Business Definition" provided in the Semantics Layer?
[ ] Trend Approval: Is the "Favorable Trend" (e.g., Up is better) signed off by the Persona?
Pillar 3: Governance (The Policy)
[ ] Decoupled Architecture: Is the math tool-agnostic? (Audit: Logic should be functional in SQL or Tableau without change).
[ ] Domain Isolation: Are physical tables prefaced with domains (e.g., HR_ or SLS_) to prevent cross-departmental record contamination?
[ ] Anti-Persistence: Is there a 1:1 map between the human-readable math and the executable code?
Pillar 4: Instruction (The Documentation)
[ ] Transparency: Does the Descriptive Math Formula explain the logic to a non-technical stakeholder?
[ ] NLP Readiness: Are logical field names singular (e.g., Customer not Customers) for AI optimization?
[ ] Rationale: Does the "Why this is important" column explain the risk of failure for each field?
Pillar 5: Data (The Five-File Artifacts)
[ ] 1. Semantics Layer: Are business conditions (e.g., Exclude returns) clearly stated?
[ ] 2. LDD: Is the Hardened Math implemented (e.g., IF SUM(ZN([Sales])) > 0)?
[ ] 3. ERD: Is the Cardinality (1:N, 1:1) verified to prevent data duplication?
[ ] 4. DD: Is the "Physical Source of Record" identified for every field?
[ ] 5. Lookup Table: Are integer codes (e.g., Segment 2) mapped to business labels (Corporate)?