Fabric and Azure
How to Migrate to Microsoft Fabric Without Losing Trust
Planning to migrate to Microsoft Fabric? See what moves, what breaks, how to phase it, and what Fabric capacity costs once you are on it.
Austin Levine
·
Aktualisiert
To migrate to Microsoft Fabric means moving your reports, semantic models, and data pipelines onto one Microsoft platform in a planned order, not all in one weekend. The safe way is to move in phases, check that the numbers still match at each step, and turn the old tools off only once the new ones are trusted. Done in the right order, it lowers cost and gives every report one place to read from.
Your finance team runs Power BI on Premium capacity, a couple of pipelines sit in Azure Data Factory, and the warehouse lives somewhere else again. Someone senior has now asked when you'll bring all of it onto one platform. The plan to migrate to Microsoft Fabric sounds simple until you open the list of what actually has to move, and you see that every report, every refresh, and every access rule is tied to something you can't break.
This is a plain guide for a Head of Data who has to give a date and then stand behind it. It covers what moves, what breaks, how to phase the work, and what Fabric capacity costs once you're living on it.
Key takeaways
Migrating to Fabric moves reports, models, and pipelines onto one platform in phases, not one weekend.
Check that the numbers still match at each step, and turn old tools off only once the new ones are trusted.
A move from Power BI to Fabric rarely breaks the reports; it breaks the plumbing around them.
The migration is done when numbers match, every table has an owner and refresh, and old tools are off.
What does it mean to migrate to Microsoft Fabric?
Microsoft Fabric puts data storage, pipelines, warehousing, and Power BI reporting under one roof, with a single storage layer called OneLake underneath. A microsoft Fabric migration means moving the pieces you run today into that one place, so a report, a pipeline, and a warehouse table all read from the same copy of the data instead of three separate ones.
For most teams the parts that move are the Power BI workspaces and datasets, the pipelines that load and refresh data, and any warehouse or lakehouse the reports sit on. What doesn't change is the way people ask for numbers. A manager still opens a report and expects one revenue figure. The point of the move is to make that figure come from one governed source, not from whichever table a given report happened to point at.
The mistake to avoid is treating this as a one-click upgrade. Fabric is a new home for your data, and moving house is a project with an order to it, not a switch you flip on a Friday.
What breaks during a Power BI to Fabric migration?
This is the part that keeps a Head of Data awake, so say it plainly. A move from Power BI to Fabric rarely breaks the reports themselves. It breaks the plumbing around them.
Here's what tends to go wrong when a team moves too fast:
Refresh schedules and gateway connections don't carry over cleanly, so a report that worked yesterday shows stale numbers the morning of a review.
Row-level security rules get rebuilt by hand, and one missed rule means a regional manager can suddenly see every region.
Workspace permissions reset, so people either lose access they need or gain access they should never have.
Reports built on DirectQuery behave differently on Direct Lake, and a few measures return the wrong total until the model is checked.
None of these are reasons to avoid the move. They're reasons to move in order and test at each step. The worst outcome is a migration that stalls half finished, with both the old Premium setup and the new Fabric setup running at once, nobody sure which report is the real one, and the monthly bill covering both.
How should you plan a project to migrate to Microsoft Fabric?
Move in phases and prove the numbers match before each phase counts as done. A workable order looks like this.
Phase | What moves | Done when |
|---|---|---|
1. Set up | OneLake, a Fabric workspace, capacity sized for a pilot | A test report runs and matches the old one |
2. Pilot | One important report and its pipeline | Three people pull the same number and agree |
3. Core | The main reports, semantic models, and refreshes | Every key metric ties out against the old setup |
4. Governance | Row-level security, owners, refresh schedules | Access and ownership are set and checked |
5. Switch off | The old Premium setup and any duplicate tools | The old tools are dark and nobody has reopened them |
The reason to phase it's confidence. After each phase you can point at a report and say the number is right, which is the only proof that matters to the people asking. The demand behind these projects is real: Microsoft Fabric passed 28,000 customers, the fastest-growing product in the company's history (Microsoft, Ignite 2025). That momentum is why the request to migrate to Microsoft Fabric is landing on so many desks this year, and why doing it in a careful order is worth the extra weeks.
What will Fabric capacity cost after you move?
Fabric runs on a capacity you rent by the hour, sized in units called F-SKUs. The bill depends on the size you pick and whether you leave it running around the clock or turn it off when nobody is working. A capacity that's too large slowly burns money, and one that's too small makes reports slow at month-end when everyone loads them at once.
The trap during a move is the overlap. For a few weeks you're paying for the old Premium capacity and the new Fabric capacity together. That's normal, but it's a reason to keep the switch-off phase short and firm. Right-sizing the Fabric capacity after the move, and turning it off outside working hours where the workload allows, is usually where the cost case for the whole project is won or lost.
Get the sizing wrong and the finance team that approved the move sees the bill go up, not down, which is the fastest way to lose their trust in the next project.
How do you know the migration is done?
The move is finished when three things are true. The numbers in Fabric match the old figures for every metric that matters. Every important table has an owner, a visible refresh schedule, and the right access rules. And the old tools are switched off, with nobody reopening them to double-check.
That last point is the real test. If people still keep the old Premium report open next to the Fabric one, the migration isn't done, it's just duplicated. The job is complete when the team trusts the new source enough to close the old one.
If you want to see where your setup stands before you commit to a date, our free Power BI Governance and Cost Scorecard walks you through the readiness and cost questions in about five minutes and gives you your top risks. If the score shows gaps, book a Fabric migration scoping call with us, and we'll map exactly what moves, in what order, and what your capacity should cost when you're done.
FAQ
How long does it take to migrate to Microsoft Fabric? For a focused set of reports, a phased move usually takes a few weeks to a couple of months. The number of metrics that disagree today and how much your access rules and refreshes have to be rebuilt are what drive the timeline, far more than the size of the data.
Can we keep Power BI when we move to Fabric? Yes. Power BI is part of Fabric, so your reports and semantic models stay. What changes is where the data underneath them lives and how it's refreshed, which is the whole point of moving in the first place.
Do we have to move everything at once? No, and you shouldn't. The safe path is to pilot one important report, prove the numbers match, then move the rest in phases and switch off the old tools last. Moving everything in one go is what causes migrations to stall half finished.
Read next: Microsoft Fabric: What It's and When to Move and Fabric AI Readiness: What Copilot Needs First.
