Skip to main content

Troubleshoot Power BI Lineage When Columns Are Renamed

Use this guide when column or field lineage in Catalog looks incorrect after you rename columns in Power BI Power Query. You'll learn typical symptoms, why renames affect lineage, which rename patterns Catalog resolves best, how to validate integrations and timing when something still looks off, and where to look for issues unrelated to column renames, such as measure-level detail.

Symptoms You Might See​

You might notice one or more of these behaviors:

  • Field lineage stops after a rename step - Table-level lineage still shows upstream warehouse tables or Dataflows, but field lineage doesn't continue through the rename.
  • Some columns trace end to end - Others stop at the semantic data set or Dataflow entity after heavy merging, splitting, or renaming.
  • Lineage was incomplete before you changed the model - You simplified rename steps and want to know what to expect after Catalog finishes another extraction cycle.

Lineage is computed using recent activity. Confirm that the asset has been refreshed during the period Catalog uses for lineage, as described in Lineage troubleshooting.

Why Renames Affect Lineage​

Catalog links Power BI fields to upstream warehouse columns by analyzing Power Query M, mashup metadata, and related admin API output together with the warehouse graph. When you rename a column, the field name in the model can differ from the name the warehouse integration ingested. Catalog needs a clear mapping in M and metadata to connect the new name back to the source column.

When that mapping is explicit and stable, column-level lineage can run end to end. When M builds rename logic indirectly or at runtime, the same logical rename can be harder to trace automatically, so column-level lineage can stay partial even when table-level links exist.

What Catalog Resolves for Renames​

Catalog improves column-level lineage for direct Table.RenameColumns usage in Power Query M, including a single mapping and multiple rename pairs in one step. In practice, that means the rename list is visible in M as a static table of old and new names that the engine can read from the step, not only the outcome after arbitrary logic runs. When generated M shows plain old-name and new-name pairs for those steps, lineage mapping back to warehouse columns is strongest.

If you use other ways to relabel fields, for example by only changing display names without a clear M-level rename step, or by using rename patterns the parser doesn't treat as direct Table.RenameColumns mappings, field lineage can remain weaker until the model uses a pattern Catalog can map.

Dynamic or Indirect Rename Logic​

Some models build the rename list at query time, for example by zipping two lists, reading from parameters, or computing which columns to rename in a step that doesn't leave a fixed pair list in the same form as a direct Table.RenameColumns call. In those cases, column-level lineage can stay incomplete because the relationship between old and new names isn't expressed the same way in the metadata Catalog ingests.

When you can change the model, use this sequence:

  1. Simplify to explicit renames for the fields you need in lineage, using Table.RenameColumns with a direct mapping in M where possible.
  2. Reduce indirection in the same query chain where Catalog must tie fields to warehouse columns.
  3. After changes, follow How to Validate Lineage After Changing Renames in this guide.

Custom wrapper functions or tenant-specific M helpers that hide Table.RenameColumns inside abstractions can fall outside the built-in patterns Catalog recognizes. If lineage is still wrong after admin settings, refresh, and a full extraction cycle, open a Support ticket with a short description of the M pattern. Do not include credentials.

How to Validate Lineage After Changing Renames​

Work through these steps in order when you have fixed or simplified rename steps and want Catalog to show updated field lineage.

  1. Confirm Power BI admin settings
    In the Power BI Admin portal, keep the same Admin API Settings that Power BI setup requires for your Catalog service principal, including options that surface detailed metadata and mashup expressions when your organization uses them for lineage.

  2. Refresh or republish affected data sets
    Refresh or republish in Power BI so the service returns M and metadata that match your changes. This matters for data sets that refresh rarely or that use DirectQuery, as described in Power BI setup.

  3. Confirm warehouse tables and columns are in Catalog
    The warehouse objects your M reads must be in scope for your warehouse integration. If Catalog doesn't have a table or column, field lineage can't attach to it. Use your warehouse integration documentation, for example Snowflake, to verify scope and sync.

  4. Allow a full Catalog extraction cycle
    Catalog only reflects new M after the next successful Power BI extraction. For Catalog-managed environments, that follows the schedule you coordinate with Catalog operations. For client-managed environments, that follows your castor-extract-powerbi schedule and upload, as in Power BI setup. The first Power BI sync can take up to 48 hours; afterward, daily extraction is typical where that schedule applies.

  5. Re-check lineage after both sides have synced
    If table-level lineage looks right but some fields still don't trace, compare those columns to the rename patterns in this guide. If everything matches supported patterns and extraction succeeded, contact Support with timestamps and asset names.

After configuration is correct and extraction runs successfully, you can see more column links than before for the same model, especially where you replaced ambiguous rename logic with direct Table.RenameColumns mappings. Large tenants still follow the same timing expectations as other Power BI ingestion.

When Lineage Stays Partial for Other Reasons​

Column renaming is one reason field lineage can stop early. These situations overlap with rename issues in the UI:

  • Multiple warehouse sources in one logical table - Catalog can't always reduce paths to a single resolved column.
  • Heavy merge, join, or append steps - These steps can obscure which input column feeds an output field.
  • Parameterized or highly dynamic connection logic - Extraction may infer less. See Power BI Dataflows in Catalog and Power BI troubleshooting in Catalog for broader troubleshooting steps.

Table-level lineage can be present while column-level lineage ends at an ambiguous transform. That is expected until M and metadata give a clear enough mapping.

Measures and DAX​

Column rename behavior in Power Query (M) is separate from how Catalog treats DAX measures in lineage. Use this section when you already validated renames and admin settings but still wonder what you should see for measures.

What Field Lineage Shows for Measures​

Catalog field lineage is built around relationships in the metadata graph. When the graph resolves the path, you can see how warehouse columns and model fields connect into measures (for example which columns participate upstream of a measure in the lineage canvas). That emphasis matches how Catalog surfaces dependency and lineage structure, not a full authoring or formula editor experience.

Full Measure Definition Text in Lineage Views​

Field lineage and lineage detail in Catalog are not a substitute for the measure editor in Power BI Desktop. They do not display the full DAX measure definition (the formula text as you see it in the formula bar or measure dialog) inside those lineage experiences.

Power BI tenant settings that enhance admin APIs with DAX and mashup expressions are still required for Catalog's Power BI integration, as documented in Power BI setup. Those settings let Microsoft's APIs return expression metadata that Catalog uses during ingestion and lineage computation. Enabling them does not add the complete measure formula to every Catalog screen, including lineage panels.

DAX Compared to Power Query M​

  • Power Query (M) - Shapes tables and columns before the semantic model. Column rename and mashup behavior in this guide apply to M-level steps such as Table.RenameColumns.
  • DAX - Defines calculated columns and measures on the loaded model. Catalog lineage can show how columns and measures connect in the graph when paths resolve. It does not replace Power BI Desktop for reading or editing full DAX text inside lineage.

When Admin Settings Look Correct but Measure Formulas Still Don't Appear​

If Enhance admin APIs responses with DAX and mashup expressions is enabled for your Catalog service principal's group, extraction succeeds, and column-level lineage behaves as you expect, absence of full measure DAX in the lineage UI is expected for how Catalog presents lineage today. That is not the same signal as a broken integration. For measure definitions, continue to use Power BI Desktop or your organization's standard tools for model authoring.

If you need help beyond this guide, contact Support with the data set or report name and what you expected to see in lineage.

What's Next?​