A semantic layer should define entities (e.g., Customer, Order), dimensions (e.g., Region, Category) and measures (e.g., Revenue, Count).

Each metric definition includes formulae, aggregation rules and filters.

Include descriptions, data types, units and relationships.

It’s helpful to distinguish between raw measures (sum of sales) and derived measures (month‑over‑month growth).

Document any assumptions and default date ranges.

Clear metadata ensures that tools and users interpret metrics consistently.

MageMetrics auto‑detects basic dimensions and measures and allows teams to author custom definitions using YAML or a UI, which are stored in version control.