Forum Discussion
Getting Measures from previous uploaded DirectQuery Report
Hi Applicable88
Unfortunately, in this scenario the behavior is expected and there is no supported way to recover the report or the measures.
When a report is built using DirectQuery to a Power BI dataset (especially a Microsoft-managed dataset such as User Metrics), the Power BI Service does not store the model or the DAX measures in a retrievable form. The PBIX file is the only place where report-level measures exist.
If the original PBIX was deleted, the measures are lost. “Copy report” only duplicates the visuals and layout; it does not expose or restore the model or the DAX. The measures will not appear in the original dataset, in a new DirectQuery report, or via any API or tool.
Power BI Service should not be treated as a backup or source-control system. For DirectQuery and thin reports, keeping the original PBIX file is mandatory.
For future work, the only safe approaches are to keep local PBIX backups or to place reusable measures in a dedicated dataset where possible.
If this post helps, then please consider Accepting it as the solution to help the other members find it more quickly
Hi Applicable88
If a report uses a DirectQuery source then it cannot be downloaded from the Power BI Service and only Save a copy is available. Once the original PBIX is deleted, custom measures cannot be recovered or reused, as they arenot exposed by the source.
Safest approach is always keep a local copy of the PBIX, document DAX measures separately (or store them in a central repo) and where possible recreate key logic in a composite or Import model so measures remain portable and recoverable.
Hi Applicable88
If a report uses a DirectQuery source then it cannot be downloaded from the Power BI Service and only Save a copy is available. Once the original PBIX is deleted, custom measures cannot be recovered or reused, as they arenot exposed by the source.
Safest approach is always keep a local copy of the PBIX, document DAX measures separately (or store them in a central repo) and where possible recreate key logic in a composite or Import model so measures remain portable and recoverable.