1,000 reports moved from Tableau to Power BI.

Where the customer stood

At Telefónica Germany, 300 employees worked with around 1,000 reports in Tableau. Reporting was to move to Power BI, but not by simply redrawing the reports there. That first needed a shared data foundation to build them on.

On top of that came extracts, data copies and several versions of the same analysis held by different teams. In her talk, Laura Ortner, Senior Licence Manager at Telefónica Germany, called it "a sprawling, uncontrolled reporting ecosystem".

It started with a question about cost: what was this estate costing the business? That quickly led to a second question. Which of it was actually needed, and what should it run on in future?

Copying the reports across one by one would have produced the same landscape under a new name. So our brief was a data platform on Microsoft Fabric. The reports would share one data estate. Permissions would be managed in one place. And Telefónica's own team would run it.

What we did

We built a data foundation on Microsoft Fabric and OneLake, then migrated the 1,000 reports to Power BI on top of it. Direct Lake connects the reports straight to the data, with no copies in between. Alongside the migration, we prepared the 300 employees for Power BI through enablement. So after the migration, reporting is run by their own people as business as usual, not by us.

We started with the estate itself. Usage data showed which reports people opened and which nobody had looked at for months. Comparing the data sources showed where several workbooks loaded the same table, each in their own way. We merged duplicates and standardised names and structures. Anything nobody used was left behind.

Then came the foundation: OneLake as shared storage with a lakehouse on top. Pipelines, semantic models and reports all use the same data. Shadow extracts that ran overnight became load processes you can trace.

Permissions are managed centrally through roles in the semantic model, workspaces and apps. Deployments run through pipelines from development to test to production and leave an audit trail. Fabric capacity is measured and managed, so the costs of the new platform cannot creep up unnoticed.

The reports moved across in waves, the most complex ones first. After that, the rest became routine. Tableau and Power BI think differently, so we translated patterns rather than formulas. Each wave was checked against the old platform and signed off by the business before we switched over. Both platforms ran side by side until the figures matched.

Telefónica's development team worked on every wave, on the real estate and on the new platform. A training programme ran from the first wave onwards.

What came out of it

After eleven months the last wave was signed off and Tableau was switched off. The licence ended, and with it the effort of running a platform nobody needed any more. Reporting carried on throughout without a single day of downtime.

What remains is one data estate in OneLake instead of many copies. Permissions sit in one place, deployments leave a trail and capacity costs are visible. The reports run in Power BI on this foundation, and the same data is ready for Copilot and for agents.

The platform lives in Telefónica's own environment, under its own licences and fully documented, and its own team runs it. Laura Ortner presented the project at SAMS Nordic 2026 in Copenhagen and SAMS Europe 2026 in Amsterdam. At TDWI Virtual II 2026 we presented it together.

What proved harder than expected

A migration of this size is largely about enablement, and only then about technology. Without the work with the 300 employees, the reports would still have arrived in Power BI. But nobody would have opened them.