The Problem We Are Solving
Maybe you inherited a Power BI dataflow that still refreshes every morning. Reports depend on it, nobody has mapped all the connections, and now Microsoft calls Dataflow Gen1 “legacy”.
That word makes the situation sound more urgent than it is.
Legacy means Gen1 is no longer receiving new feature investment. It does not mean every working Gen1 dataflow must be replaced immediately. Microsoft says existing Gen1 workloads can continue to run, and Pro or Premium Per User customers can continue using Gen1 while its guidance for those scenarios evolves.
The useful decision is not “Is Gen2 newer?” It is:
Should we keep this Gen1 dataflow, test Gen2 alongside it, or plan a controlled migration?
The Short Answer
Do not migrate a working dataflow only because it has acquired a legacy label.
- Keep it for now when it is stable, understood and still fits your licensing.
- Test Gen2 separately when you have Fabric access and want to learn how the same queries behave without disturbing a live report.
- Plan a migration when the dataflow is changing, you need Gen2 capabilities, or the existing workflow is becoming harder to support.
Whichever route you choose, inventory the reports, semantic models, Excel files and other dataflows that depend on the current item before you touch it. Microsoft puts that inventory step first in its migration guidance for a reason: copying the Power Query logic is usually easier than finding everything that will be affected afterwards.
Choose Your Route
| Your situation | Sensible next move | What to do now |
|---|---|---|
| Your Gen1 dataflow is stable, business-critical and poorly documented | Keep Gen1 while you map it | Record its owner, sources, schedule and every known downstream report or workbook. Do not use a live migration to discover its dependencies. |
| You use Pro or PPU and do not have a Fabric-capacity workspace | Keep Gen1 | Microsoft still supports this route. Ask what Fabric access, if any, your organisation plans to provide before designing a migration. |
| You have Fabric access and are building something new or substantially changing the flow | Start with Gen2 | Build or test the new workload in Gen2 rather than extending a legacy design by default. |
| You have Fabric access but are unsure whether the existing flow will move cleanly | Run a side-by-side test | Create a separate Gen2 copy, leave Gen1 running and compare the results before changing any consumers. |
| You need Gen2 destinations, improved monitoring or Fabric deployment features | Plan a controlled migration | Inventory dependencies, confirm capacity and permissions, choose a migration method, then test the replacement as a separate item. |
| Someone wants an immediate migration because “Gen1 is being switched off” | Pause and verify the claim | Microsoft’s current guidance describes Gen1 as supported but legacy. Build a plan from your real constraints, not an assumed shutdown date. |
First, Make a One-Page Inventory
You do not need an enterprise architecture exercise. You need enough information to avoid breaking the monthly report.
Copy this table into Excel or a work note and complete one row for each dataflow:
| Check | What to record |
|---|---|
| Dataflow | Name, workspace and current owner |
| Licence and capacity | Whether the workspace is Pro, PPU, Premium or assigned to Fabric capacity |
| Sources | Files, folders, databases, gateways and credentials used |
| Refresh | Schedule, time zone, typical duration, failures and incremental-refresh settings |
| Downstream use | Power BI semantic models, reports, Excel workbooks, linked dataflows, Power Apps or other consumers |
| Business checks | Row counts, key totals and KPIs that must match after migration |
| Change owner | Who can update connections, schedules and downstream reports |
| Rollback | How long the original Gen1 item will remain available during testing |
If you cannot identify the owner or downstream use, that is not a reason to migrate faster. It is the first risk to fix.
Confirm That You Can Test Gen2
For the documented Save as Dataflow Gen2 migration route, Microsoft lists three prerequisites:
- The target workspace must use Fabric capacity of F2 or higher, or a Fabric trial.
- The tenant setting Users can create Fabric items must be enabled for you or your group.
- You need the Contributor workspace role or higher in the target workspace.
Power BI Pro and PPU workspaces on their own do not support Dataflow Gen2.
If you do not know the answers, send this to your Fabric or Power BI administrator:
I am assessing whether an existing Power BI Dataflow Gen1 should move to Dataflow Gen2. Could you confirm whether the target workspace is assigned to Fabric capacity (F2 or higher) or a trial, whether I am allowed to create Fabric items, and whether I have Contributor access or higher? I am planning a side-by-side test and will not replace the live Gen1 item during the assessment.
Pick the Least Disruptive Migration Method
Microsoft documents three ways to move Gen1 queries into Gen2:
- export the Gen1 queries as a Power Query template and import them
- copy and paste the queries in Power Query
- use Save as Dataflow Gen2 to create a new Gen2 item
For an inherited dataflow, Save As is a useful starting point because it creates a separate Gen2 item that you can review. It is not a complete cutover.
Two settings need particular attention:
- Scheduled refresh settings are not copied by Save As.
- Incremental-refresh settings are not copied when saving a Gen1 dataflow as Gen2.
Treat the new item as an untested replacement until you have recreated the required settings and checked every downstream connection.
Test Before You Switch Anything
A successful refresh only proves that the new dataflow ran. It does not prove that the business report is still right.
Before moving a report or semantic model to Gen2, check:
- every expected query is present
- source credentials and gateway access work
- the refresh completes at an acceptable time
- row counts are within the expected range
- important dates, IDs and null values still behave correctly
- the totals and KPIs people challenge in meetings match the approved baseline
- scheduled and incremental-refresh behaviour has been recreated where needed
- a test report or semantic model can consume the new output
- the original Gen1 item can remain available until the replacement has passed an agreed test period
Move consumers in controlled groups. Start with a test report, not the report used in tomorrow morning’s leadership meeting.
Watch the Destination Settings
Gen2 can send each query to a destination such as a Fabric Lakehouse, Warehouse or another supported store. That flexibility introduces a setting worth checking carefully.
For new Lakehouse and Azure SQL destination tables, automatic settings use replace behaviour and can drop and recreate the table during refresh. Microsoft warns that this might remove relationships or measures previously added to that table.
If a downstream semantic model depends on relationships or measures attached to the destination table, review the destination and schema settings before the first production refresh. Microsoft documents fixed-schema settings as the option that keeps existing relationships and measures intact, although a refresh or publish will fail when the query schema no longer matches. That is a trade-off to test, not a switch to change blindly.
What Still Transfers From the Original Session
The useful Power Query work has not become obsolete.
Microsoft’s supported migration methods are based on moving or reusing the existing queries. Your transformations may transfer, but the surrounding operating setup still needs attention: capacity, permissions, schedules, incremental refresh, destinations and downstream consumers.
That is the practical 2026 update:
Keep the working logic where it is safe, test Gen2 where it is useful, and migrate only after you know what depends on the current dataflow.
Evidence Notes
Microsoft documentation supports these decisions:
- Gen1 is supported but in a legacy state and is not receiving new feature investment. Existing workloads can continue while teams plan their transition.
- Pro and PPU customers can continue using Gen1. Gen2 requires an appropriate capacity-backed workspace or trial.
- Microsoft recommends taking an inventory of dataflows and dependent items before migration.
- Microsoft documents Power Query template, copy-and-paste and Save As migration routes.
- Save As does not copy scheduled-refresh or Gen1 incremental-refresh settings.
- Automatic destination settings can drop and recreate some destination tables, which may remove relationships or measures.
The documentation does not prove that your migration will preserve business totals, run at the same cost, finish in the same time, or leave every downstream report untouched. Only a side-by-side test against your own sources, refresh pattern and agreed KPIs can establish that.
Your Next Move
Choose one important Gen1 dataflow and complete the inventory table before anyone changes it.
At the end of that check, record one decision:
- Keep Gen1 and review again on a named date
- Run a side-by-side Gen2 test
- Plan migration with an owner, test period and rollback route
If you cannot yet name the downstream consumers or the person responsible for the change, the decision is inventory first.
Source note
Briefing published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "Should You Keep or Migrate Your Power BI Dataflow in 2026?", Collab365 Spaces. 7 sources referenced.