Power BI Collected Data

This page lists what Sifflet imports from Power BI once you connect it.

Imported Assets

Sifflet imports the following assets from each selected workspace:

Reports
Dashboards
Tiles: dashboard tiles created directly on a dashboard. Tiles pinned from a report aren't imported as separate assets: the dashboard is linked to the report instead.
Semantic models (formerly datasets), shown with the Dataset type in Sifflet
Semantic model tables, with their columns, shown with the Table Ref type in Sifflet (or Model when the table is defined by a native SQL query)

The Sifflet Insights browser extension supports the following Power BI asset types:

Reports
Dashboards

Collected Metadata

Sifflet reads all metadata through the Power BI metadata scanning APIs:

Asset type	Metadata
Report	Name, description, last modified by, semantic model it uses
Dashboard	Name, tiles and reports it contains
Tile	Title, semantic model or report it comes from
Semantic model	Name, tables, upstream semantic models
Semantic model table	Name, description, columns (names and data types), Power Query (M) source expression

Sifflet doesn't collect Power BI usage statistics (views, users).

Lineage

Sifflet builds Power BI lineage from the metadata scan results:

Within Power BI: semantic model tables → semantic models → reports and tiles → dashboards. When a semantic model reads from another semantic model, the report is linked to the upmost semantic model.
To your data warehouse: Sifflet parses the Power Query (M) source expression of each semantic model table to find the warehouse table it reads, and links it to the matching table in your Sifflet sources. Supported platforms are Snowflake, BigQuery, Redshift, Databricks, and Synapse.
Native queries: when a semantic model table is defined by a native SQL query (Value.NativeQuery) on Snowflake, BigQuery, or Redshift, Sifflet parses the SQL query and links the table to the warehouse tables it reads.

When a source expression uses semantic model parameters (for example a parameterized database name), Sifflet resolves them only if the service principal has the Member role on the workspace. See Permissions Required.

Data Freshness

Sifflet refreshes Power BI metadata on the source's schedule (weekly by default for BI tools). See Integrations Management for scheduling and manual refreshes.

Limitations

Lineage to the data warehouse is only built for semantic model tables that read from Snowflake, BigQuery, Redshift, Databricks, or Synapse. Tables from other sources (files, other databases, dataflows) are imported without upstream lineage.
Native SQL queries are parsed for Snowflake, BigQuery, and Redshift only.

Data Exposure

Sifflet only reads Power BI metadata (names, descriptions, structure, source expressions). It doesn't read or store the data of your reports or semantic models. See Security for how Sifflet stores and protects metadata.