Power BI write back, finally native: translytical task flows explained

For years, "can users update the data from the report?" ended every Power BI conversation with no, export to Excel. Translytical task flows change the answer: a button in the report calls a Fabric user data function, which writes to the source — annotations, status updates, approvals, discount changes, all without leaving the page.

The moving parts

Three pieces make a flow:

• A user data function in Fabric — your code (Python), with native connection management for Fabric SQL databases, warehouses and lakehouse files. Microsoft recommends a SQL database for most write-back, since it handles mixed read/write load well. The function must return a string — that string becomes the message the report shows.
• A data function button on the report page, its action type set to Data function, picking your function from the OneLake catalog.
• Input controls — slicers, selected rows or measures — bound to the function's parameters.

The semantic model can be DirectQuery (updates appear live), Direct Lake, or even Import — with Import the change lands in the source but waits for the next refresh to show.

Wire the button

1. Add a button, open Action, set Type to Data function, and choose the function via the fx picker.
2. Each function parameter appears below: bind one to a slicer on the page, another to a data field or measure (with a summarisation such as Maximum when rows aggregate).
3. The button stays disabled until every parameter has input — clicking a table row or filling a slicer enables it, which doubles as your wiring test.
4. Style the Loading state: data function buttons get an extra button state with a spinner, for text like "Submitting…" while the function runs.
5. Optionally turn on Auto clear so input slicers empty themselves after a successful call.

Collect input with the input slicer

The input slicer (the renamed text slicer, now generally available) has two personalities: with a data column it filters; without one it's a pure input control — a text box whose value exists to be passed into your function. That's the pattern for comments, new values and annotations. A button slicer beside it covers pick-one choices such as a status.

Pause animation

1. 1Reader selects a row and types a value
2. 2The button enables
3. 3The user data function validates and writes
4. 4The report shows the function's message

Harden the function before anyone clicks

Microsoft's own best-practice list is the difference between a demo and something you'd let a department use:

Beyond plain write-back, the same machinery powers status workflows with history and Teams notifications, and in-report alerts where notifications are just rows your function writes and the report displays. Microsoft ships tutorials for both.

Sources

• Understand translytical task flows — Microsoft Learn
• Create a data function button in Power BI — Microsoft Learn
• Tutorial: create a translytical task flow — Microsoft Learn
• Create and use an input slicer — Microsoft Learn