answering-natural-language-questions-with-dbt

To answer business questions with dbt's semantic layer, query existing metrics and dimensions to retrieve direct insights from your data warehouse. The system prioritizes the semantic layer for analytics requests like total sales or customer counts.

Query data warehouse metrics by falling back to ad-hoc SQL or compiled SQL when semantic layer coverage is incomplete. The system automatically transitions from semantic layer queries to compiled SQL to generate executable queries for missing metrics.

Yes, you can use dbt to analyze KPIs and ad-hoc data inquiries such as regional performance or quarterly trends. The system answers analytics questions by querying marts and models directly from the data warehouse.

The best way to get direct insights using dbt is applying a decision flow that prioritizes the semantic layer, then compiled SQL, model discovery, and manifest analysis. This approach ensures executable queries retrieve accurate analytics.

You should not use dbt semantic layer queries for validating, testing, or building dbt models during development. This approach is strictly designed for answering analytics, metrics, KPIs, and data inquiries from existing data warehouse models.