Semantic Layer

A layer that centralizes governed definitions of metrics, dimensions, filters, and query relationships over data. It lets dashboards, APIs, and agents ask the same business question and reach the same calculation, rather than having every consumer reimplement rules in SQL or application code.

Metric consistency does not automatically resolve the identity of the entities entering a calculation. If “active customer” joins tables that represent different populations or poorly reconciled keys, the result can be reproducible and still be wrong. A semantic layer needs contract tests and ownership and, when multiple systems are involved, an ontology or another explicit entity-resolution mechanism.