dbt Cloud is widely used as the SQL transformation layer in modern data stacks. dbt models produce the tables and views that Power BI datasets query. Integrating dbt Cloud monitoring with Power BI monitoring closes the gap between the transformation layer and the reporting layer.
What dbt Cloud integration covers
Job run monitoring: dbt Cloud jobs run transformations on a schedule or triggered by CI/CD. The dbt Cloud API returns job run status, run duration, and whether any models within the run failed. A job can complete with a failed status (at least one model failed) or a partial run state (the job was cancelled before finishing).
Model-level error detail: When a dbt job fails, the error is typically in a specific model. The dbt Cloud run steps API returns which model failed and the SQL error message. This is the root cause information — not just "the job failed" but "model compute_margins failed with error column not found: discount_pct." MetricSign includes this as a root_cause_hint in the incident.
Manifest lineage parsing: dbt generates a manifest on each run. MetricSign parses this manifest to build a model-to-Power BI dataset lineage map. By matching dbt model output tables against Power BI datasource configurations, it links dbt job failures to the downstream datasets they affect.
Schema change detection: When the manifest changes between runs — because a developer modified a model's SQL and changed its column structure — MetricSign detects the change and surfaces it as a potential risk to downstream Power BI measures and relationships.
Setting up dbt Cloud integration
Configuration requires a dbt Cloud service token with the jobs:read and runs:read permissions, the dbt Cloud account ID, and the project ID. MetricSign extracts these from the dbt Cloud URL if provided, and uses the token to poll the API for job run status after each run completes.