Power BI

BI Tools

Catalog your Power BI workspaces, semantic models, reports, and dashboards - and trace every table and column back to the warehouse objects that power them.

Features

Feature Name	Supported	Notes
Catalog Discovery	✓	Workspaces, Apps, Datasets, Tables, Columns, Measures, Reports, Pages, Dashboards, Tiles
Table-level Lineage	✓	Traces dataset tables to upstream warehouse tables by parsing M-Query (Power Query) expressions
Column-level Lineage	✓	Links dataset columns to warehouse columns, including through transformations that preserve column names
Multi-connection Support	✓	Distinguishes separate warehouse connections within the same Power BI tenant
Descriptions	✓	Asset descriptions from Power BI are surfaced in the Catalog
Endorsement Tags	✗	Not yet surfaced
Usage Statistics	✓	View counts and last-viewed timestamps on reports and dashboards

What gets cataloged

Matia syncs the following Power BI objects into your catalog:

Power BI Object	Cataloged as	Notes
Workspace	Asset	The root container for everything else
App	Asset	A distribution bundle. Reports and dashboards nest under the app that publishes them, and directly under the workspace otherwise
Dataset (semantic model)	Asset	The data-modeling layer; contains tables
Table	Asset	Belongs to a dataset; carries columns and measures
Column	Field	Typed from the dataset schema
Measure	Field	Sits alongside columns as a table field, and carries its DAX expression
Report	Asset	Bound to exactly one dataset; contains pages
Page	Field	A leaf under its report
Dashboard	Asset	A pinboard of tiles
Tile	Asset	Each tile gets its own asset page, where its lineage lives

Lineage

Matia derives the lineage by parsing the M-Query expression behind each table in a dataset, combined with metadata from the Power BI Admin API.

Column-level lineage

Column-level lineage narrows this further, linking fields directly:

This resolves when a table reads its source directly - either a navigation into a warehouse table, or a native SQL query via Value.NativeQuery, whose SQL Matia parses to map each column to the warehouse columns it selects.

Column-level lineage also survives name-preserving transformations. Dropping, selecting, reordering, filtering, or sorting leaves the remaining columns corresponding to the warehouse columns they came from, so the links still hold:

Table.RemoveColumns · Table.SelectColumns · Table.ReorderColumns · Table.SelectRows · Table.Distinct · Table.Sort · Table.Buffer

Name-changing transformations break that correspondence - a renamed, unpivoted, or combined column can no longer be traced to a single warehouse column by name. There, Matia keeps the table-level link and drops the column-level one rather than guessing:

Table.RenameColumns · Table.Combine · Table.Unpivot · Table.Pivot · Table.AddColumn · Table.Group

Column-level lineage covers columns read from the warehouse. Fields computed in the semantic model - calculated columns and measures - are cataloged with their expressions, but have no upstream warehouse column to point at.

Usage Statistics

Every cataloged report and dashboard shows how it is actually being used - Matia presents which users viewed the report/dashboard, when they viewed them and how many times.

Usage is sourced from the Power BI activity log via the Admin API, which is why read-only Admin API access is a prerequisite. Usage covers reports and dashboards, datasets and tables have no view activity of their own.

Usage history starts accumulating when you connect. Matia will show the usage statistics from the day you connect the asset and forward.

Activity in workspaces you have not connected to Matia is not ingested.

Limitations

Name-changing transformations lose column-level lineage. A table reshaped by Table.RenameColumns, Table.Combine, Table.Unpivot, Table.Pivot, Table.AddColumn, or Table.Group still links to every warehouse table it reads at table level, but produces no column-level edges. Name-preserving transformations keep theirs.
Fields computed in the semantic model - calculated columns and measures - are cataloged with their expressions, but have no column-level lineage, since there is no upstream warehouse column to link to.
Missing column metadata. Some tables return no column metadata due to permissions or connector-specific limits. Table-level lineage still appears; column-level does not.
Cross-workspace datasets. If a report is bound to a dataset published in a different workspace, that workspace must also have the Matia app added as Contributor. Otherwise lineage for that report will be incomplete.
Paginated reports. PaginatedReport is a report variant with no pages - the Power BI API does not expose them for this report type, so none are cataloged.
Image and text tiles. A tile links back to a report only when it was pinned from a report page. Image and text tiles have no such edge.
Endorsement tags are not yet surfaced in the Catalog.

Next Steps

Setup Guide — Connect Power BI
Release Notes — See recent changes