Power BI & Data

Power BI Dataflows in 2026: Gen1, Gen2 and When to Use Them

Understand what Power BI dataflows do, the current Gen1 legacy status, when Fabric Dataflow Gen2 fits, and what to check before centralising Power Query work.

Collab365 Team · 30 March 2026 · Updated 24 August 2026 · 2 min read

Dataflows move Power Query transformations out of individual reports and into a cloud-managed preparation layer. That can reduce duplicated query logic, but it also creates a shared dependency that needs ownership, refresh monitoring and change control.

The product choice has changed. Microsoft now describes Power BI Dataflow Gen1 as legacy with no new feature investment. For Premium customers with Fabric access, Microsoft recommends evaluating Dataflow Gen2. Pro and Premium Per User customers can continue to use Gen1 while the guidance for those scenarios evolves.

What a dataflow does

A dataflow connects to sources, applies Power Query transformations and writes resulting tables to a cloud destination. Reports and other services can then consume those prepared tables instead of each report repeating the same extraction and cleaning steps.

This is useful when several reports need the same customer, product or calendar preparation. It is less useful when a transformation belongs to one small report and centralising it would add more operational work than it removes.

Gen1 or Gen2?

Treat this as an architecture decision, not a simple upgrade button.

Dataflow Gen1 remains relevant in existing Power BI environments, particularly where the team has Pro or Premium Per User licences and a working operating model. Its legacy status means you should avoid building a new long-term strategy without considering the alternative.

Dataflow Gen2 is part of Fabric Data Factory. It supports multiple destinations and a different operating model. Capacity, workspace configuration and licence requirements matter. Microsoft provides migration guidance and a feature comparison because not every Gen1 design maps identically.

Check:

  • the licence and Fabric capacity available to creators and consumers
  • supported source connectors and gateways
  • destination requirements
  • refresh frequency and concurrency
  • incremental-refresh support for the chosen route
  • deployment and source-control needs
  • monitoring, retry and ownership

Do not accept a generic claim that Gen2 is faster or cheaper for your workload. Test representative queries and refreshes in your environment.

A safe adoption pattern

  1. Identify query logic duplicated in at least two maintained reports.
  2. Agree the table contract, owner and change process.
  3. Build the dataflow in a non-production workspace.
  4. Reconcile row counts, keys, types and null behaviour against the current reports.
  5. Connect one report and compare its measures and refresh history.
  6. Document gateway, credentials and failure notifications.
  7. Migrate other consumers only after the first route is stable.

Centralising a bad query gives the organisation one authoritative bad query. The quality gate still matters.

Dataflows do not replace the semantic model

The dataflow prepares and stores tables. The semantic model still owns relationships, measures, model-level security and the analytical view consumed by reports. Keep those responsibilities explicit.

Sources

Work through the operating model

The Microsoft 365 Report Builders Space is for practical reporting decisions such as shared transformations, refresh ownership and migration boundaries.