Methodology
What is a metric contract?
A metric contract is a reviewable definition of a business KPI: the question it answers, how it is calculated, where it comes from, who owns it, how it is validated, and how it should not be used.
Why metrics break
Metrics fail socially before they fail mathematically. Two dashboards can both be "correct" while answering different questions — different grains, filters, refund policies, or attribution windows.
Without a contract, debates become opinion contests. With a contract, disagreements become change requests against a shared artifact.
How to validate a metric
- Confirm owner, business question, and formula exist.
- Lock grain and default filters to prevent silent scope drift.
- Require at least one validation rule with an owner action.
- For rates/ratios, define numerator and denominator explicitly.
- Document limitations and incorrect usage examples.
How to read the maturity score
The score is explainable and capped at 100. It rewards business clarity, formula/fields, source/grain, validations, limitations/examples, and ownership/maintenance. A high score does not mean the SQL is production-perfect — it means the contract is governable.
What this product does not do
- It does not execute SQL against your warehouse.
- It does not replace dbt semantic layers in the MVP.
- It does not manage permissions or multi-user workflows.
- It does not auto-generate metrics with AI.
Full write-up in docs/metric-contract-methodology.md. Explore demos on the examples page.