Analytical Explainability
Lineage: Tracing a Number Back to Its Sources
Lineage makes every number auditable by tracing it to sources.
TL;DR
- • Lineage shows where a number comes from.
- • AI explanations rely on clear lineage.
The problem (layman)
- • Users can’t tell which tables or transformations produced a number.
- • AI answers lack auditability.
Why it matters
- • Lineage builds trust and enables debugging.
- • It supports governance and compliance.
Symptoms
- • Analysts spend time tracing calculations manually.
- • AI explanations are vague or untraceable.
Root causes
- • No metadata on measure lineage.
- • Complex transformations without documentation.
What good looks like
- • Measures list source tables and columns.
- • Lineage is visible in reports and AI responses.
How to fix (steps)
- • Add lineage annotations to key measures.
- • Document transformation steps.
- • Expose lineage in answer summaries.
Pitfalls
- • Relying on external documentation only.
- • Skipping lineage for derived measures.
Checklist
- • Lineage documented for top KPIs.
- • Transformation steps captured.
- • Lineage surfaced to users.