Tableau to Power BI Migration: The Complete 2026 Guide
A Tableau to Power BI migration is a rebuild, not a file conversion. Power BI Desktop cannot open a .twb or .twbx file, so each workbook is recreated as a semantic model plus a report, and every calculated field is rewritten in DAX. The saving is real but it comes mostly from author seats, not viewer seats: Tableau Creator lists at USD 75 per user per month against Power BI Pro at USD 14, both list prices checked 21 August 2026. Budget 4 to 16 hours per report you keep, and expect to keep far fewer reports than you have now.
Why teams move
Four reasons come up repeatedly, and only the first is about price.
Author licence cost. A team of eight analysts on Tableau Creator seats costs USD 600 a month at list. The same eight on Power BI Pro cost USD 112.
You already pay Microsoft. If you run Microsoft 365 and Entra ID, Power BI inherits your groups, conditional access, sensitivity labels and Teams distribution with no extra identity work. Microsoft 365 E5 includes a Power BI Pro licence per user, so some organisations discover they are already paying for it.
Consolidation onto Fabric. Power BI is now the reporting layer inside Microsoft Fabric, so the same platform covers ingestion, storage and reporting. If you want that, read Microsoft Fabric vs Power BI before you decide how much of Fabric you need.
Excel. Power BI models connect to Excel through Analyze in Excel and live PivotTables. Finance teams that have quietly been exporting Tableau crosstabs to Excel for years stop doing that.
Three reasons not to move: your authors work on Macs (Power BI Desktop is Windows only), you have heavy Tableau Prep flows that would need rebuilding in Power Query or Fabric pipelines, or you sell embedded Tableau dashboards to customers under a contract you cannot exit cheaply.
The licence maths
List prices, checked 21 August 2026. Tableau publishes its list prices in USD. Microsoft publishes an INR price list for India.
| Role | Tableau Cloud Standard | Tableau Cloud Enterprise | Power BI (US) | Power BI (India) |
|---|---|---|---|---|
| Author / builder | Creator, USD 75 /user/mo | Creator, USD 115 /user/mo | Pro, USD 14 /user/mo | Pro, ₹1,165 /user/mo |
| Power user | Explorer, USD 42 /user/mo | Explorer, USD 70 /user/mo | Pro, USD 14 /user/mo | Pro, ₹1,165 /user/mo |
| Read-only consumer | Viewer, USD 15 /user/mo | Viewer, USD 35 /user/mo | Pro, USD 14 /user/mo | Pro, ₹1,165 /user/mo |
| Large models, 48 refreshes a day | No equivalent tier | No equivalent tier | Premium Per User, USD 24 /user/mo | Premium Per User, ₹1,995 /user/mo |
All figures are annual billing. Indian prices exclude GST. Tableau prices come from tableau.com/pricing and Power BI prices from the Microsoft Power BI pricing page.
Work a real example. Five authors and 100 read-only viewers:
- Tableau Cloud Standard: (5 x 75) + (100 x 15) = USD 1,875 a month.
- Power BI Pro: 105 x 14 = USD 1,470 a month.
That is a 22% saving, not the 80% the headline Creator price implies, because Power BI charges viewers almost the same as Tableau does. The picture changes when content sits on a Fabric F64 capacity or larger. On an F64 or larger capacity, a user with only a free licence and the viewer role on the workspace can open Power BI content, which Microsoft states in its Fabric licensing documentation. Only the five authors need Pro. Whether that beats 100 Viewer seats depends on the capacity price, which we break down in Microsoft Fabric pricing explained.
The rule of thumb: below roughly 100 viewers, per-user Pro licences win. Above that, run the F64 comparison.
Concept mapping
Most migration confusion comes from hunting for a one-to-one equivalent that does not exist. This is the map.
| Tableau | Power BI |
|---|---|
| Workbook (.twb / .twbx) | .pbix file, or a semantic model plus a separate report |
| Worksheet | A single visual on a report page |
| Dashboard | A report page |
| Story | A set of report pages, or bookmarks |
| Data source (.tds) | Semantic model |
| Extract (.hyper) | Import mode model in VertiPaq |
| Live connection | DirectQuery |
| Calculated field | DAX measure or calculated column |
| LOD expression (FIXED) | CALCULATE with ALLEXCEPT, or a summarised table |
| LOD expression (INCLUDE / EXCLUDE) | No direct equivalent, needs redesign |
| Table calculation | DAX window functions, or a measure over a filtered table |
| Parameter | What-if parameter, or a disconnected table |
| Set / group | Calculation group, calculated column, or a grouping table |
| Tableau Prep flow | Power Query, Dataflow Gen2, or a Fabric pipeline |
| Tableau Server / Cloud site | Power BI workspace |
| Project | Workspace or app |
| Ask Data | Q&A visual |
The two rows that cost the most time are LOD expressions and table calculations. DAX evaluates using filter context and row context, and a measure has no idea which dimensions the visual is displaying. So an INCLUDE or EXCLUDE expression that changes behaviour depending on what the user drags onto the view has no static DAX version. Those get rebuilt as a different design, usually by adding the grain to the model instead of computing it at query time.
What does not convert
Microsoft's own guidance says it "isn't possible to completely automate a Power BI conversion process end-to-end". Assume manual work for all of the following:
- Custom SQL inside Tableau data sources. Move it into a view in the source database, or into Power Query. Leaving custom SQL in place kills query folding and makes refreshes slow.
- Row-level security. Tableau user filters and entitlement tables map to Power BI RLS roles with DAX filter expressions. The logic transfers, the syntax does not.
- Dashboard actions. Filter actions, highlight actions and URL actions become a mix of cross-filtering, drillthrough pages and button actions. Behaviour will differ, so demo it to users before you promise parity.
- Custom colour palettes and fonts. Rebuild once as a Power BI theme JSON file and apply it everywhere. Do this in week one, not at the end.
- Extension-based visuals. Check the Power BI visuals marketplace early. If a Tableau extension has no equivalent, decide whether to drop the view or commission a custom visual.
- Data-driven alerts. Tableau alerts become Power BI alerts on dashboard tiles, or Fabric Activator rules. The trigger conditions available are not the same, so review each alert with the person who receives it.
Third-party accelerators sold on the Microsoft marketplace will lift field lists, layout skeletons and simple calculations. Treat their output as a first draft that a person still has to validate row by row.
The migration, step by step
Microsoft publishes a five-stage migration framework in its Power BI migration guidance. This is that framework applied to a Tableau estate.
1. Inventory everything, then pull the audit log. List every workbook, every data source and every published extract. Then pull Tableau's usage data: views per report per month, unique consumers per report, and last-viewed date. This step decides your budget, because usage data is the only defensible way to argue that a report should not be rebuilt.
2. Cut the list. Sort by usage and sponsor. Kill anything unused. Merge near-duplicates. A report that answers the same question as another report with a different filter is one report with a slicer. Every workbook you do not rebuild saves 4 to 16 hours, so this step is worth more than any tooling decision you make later.
3. Set up the tenant before you build anything. Power BI tenant settings, workspace naming and roles, Entra security groups, an on-premises data gateway cluster if you have local sources, a deployment pipeline for dev and test and production, and your theme file. Decide now whether you are on Pro licences or a Fabric capacity, because it changes how you distribute content.
4. Design the semantic models, not the reports. This is the step Tableau teams skip and regret. Tableau lets each workbook carry its own data source, so estates accumulate dozens of them. Power BI works best with a small number of certified shared models that many reports connect to. Aim for one model per business subject area: sales, finance, operations. Build them in star schema with a proper date table.
5. Build a proof of concept on your hardest report. Pick the workbook with the worst LOD expressions and the most demanding user. If you can rebuild that one and get sign-off, the rest is throughput. If you cannot, you have found your real problem in week three rather than week eleven.
6. Rebuild, validate against Tableau numbers, and get written sign-off. For every report, run the same filters in both tools and compare totals cell by cell. Do not skip this. Discrepancies almost always trace back to a default aggregation, a many-to-many relationship, or a date table that is not marked as a date table. Record sign-off per report from the business owner, not from IT.
7. Deploy, train, then switch Tableau to read-only. Run both platforms in parallel for a defined window, usually four to six weeks, with a fixed end date published up front. Set the Tableau site to read-only on that date so no new work lands there. Train report consumers separately from report authors, because they need completely different things.
Time and budget
Effort per report, once the models are built:
| Report type | Build and test |
|---|---|
| Simple KPI or list view, existing model | 2 to 4 hours |
| Standard multi-visual dashboard | 6 to 10 hours |
| Heavy LOD or table-calculation logic | 12 to 24 hours |
| Report needing a new semantic model | Add 16 to 40 hours for the model |
Add two to four weeks up front for steps 1 to 4, and two weeks at the end for training and parallel running. A 40-report estate with three or four shared models typically runs 8 to 14 weeks with one or two people on it.
Work backwards from your Tableau renewal date. Tableau seats are sold on annual terms, so a migration that finishes a month after renewal costs you a full extra year of Creator and Viewer seats. Start the inventory at least five months before the renewal date, and if the dates cannot be made to work, renew for the shortest term your reseller will agree to rather than rushing the rebuild.
On external cost: Maxwize charges ₹5,000 an hour in India, and data analytics projects internationally run USD 1,000 to USD 20,000 depending on scope. A small migration (under 15 reports, one model, no gateway work) sits at the bottom of that band. A multi-department move with Fabric in the picture sits at the top. If you are comparing quotes, how to hire a Power BI consultant lists the questions that separate people who have done this from people who have watched a video about it.
Four mistakes that sink these projects
Recreating dashboards pixel for pixel. You are paying to rebuild a report someone designed under Tableau's constraints in 2019. Ask what question the report answers and what decision follows, then design for that. Microsoft's migration guidance gives the same advice.
Skipping the semantic model layer. One .pbix per report with its own embedded model gives you the exact sprawl you left Tableau to escape, plus refresh costs multiplied by the number of reports.
DirectQuery on everything because Tableau live connections worked fine. Tableau's live connections and Power BI's DirectQuery behave differently under load. Import mode is the default answer. Use DirectQuery when the data genuinely must be current to the minute or the model is too large to import.
No end date on the parallel run. Without a hard cutover date, both platforms stay live, you pay for both, and half the users never move. Publish the date in week one.
FAQ
Can Tableau workbooks be converted to Power BI automatically?
Not fully. Microsoft's migration guidance states that it is not possible to completely automate a Power BI conversion process end to end. Third-party accelerators can lift layout, field lists and simple calculations, but every data model, LOD expression and table calculation needs a human to check it.
How long does a Tableau to Power BI migration take?
Plan 4 to 16 hours of build and test time per report you decide to keep, plus two to four weeks up front for inventory, licensing and data model design. A 40-report estate usually lands in the 8 to 14 week range with one or two people on it.
Is Power BI cheaper than Tableau?
For authors, yes by a wide margin. As of August 2026, Tableau Creator lists at USD 75 per user per month and Power BI Pro at USD 14. For viewers the gap is small (Tableau Viewer USD 15 versus Power BI Pro USD 14) until you put content on a Fabric F64 capacity or larger, where viewers need no paid per-user licence.
Do I need Microsoft Fabric to move off Tableau?
No. Power BI Pro licences plus the shared capacity in the Power BI service are enough for most small and mid-sized estates. Fabric F SKUs matter when you have hundreds of viewers, large models, or you also want the lakehouse and pipeline tooling.
What happens to Tableau LOD expressions in Power BI?
They get rewritten in DAX. FIXED expressions usually map to CALCULATE with ALLEXCEPT or a summarised table. INCLUDE and EXCLUDE have no direct equivalent because a DAX measure does not know which dimensions a visual is showing, so those need redesign rather than translation.
Where Maxwize fits
Maxwize plans and runs Tableau to Power BI migrations for small and mid-sized businesses in India, the UAE and the US, including the semantic model design that decides whether the result is maintainable. If you want a scoped estimate, send us your Tableau inventory and usage export and we will tell you which reports are worth moving. See our Power BI services or get in touch.