BI Migration

A switch of BI tools usually has an outside trigger: license costs, an end of maintenance, too many parallel systems, or the wish for a solution that fits the rest of the IT landscape better. The changeover then touches reports, data models and habits all at once.

We guide such switches, most often from Qlik, Tableau or SAP BusinessObjects to Power BI and from Power BI Premium to Fabric capacities. Other directions come up too, for instance to Tableau, Looker or the BI features of a data platform. The examples on this page target Power BI because that is the usual case. A complete rebuild is not the goal. The new system should answer the business teams’ questions better than the old one did.

Not every report moves

It starts with an inventory: reports, data sources, calculations, permissions, and who builds and who uses them. Many systems log access, and that makes a usage analysis possible. It separates the reports opened every week from those nobody has looked at in years.

Each report then gets one of four decisions. It moves unchanged, it is rebuilt because the data model or presentation should improve, it is merged with similar reports, or it is switched off. Copying every old report instead carries its weaknesses along too. What counts is which decision a report should support, not what it used to look like.

The logic has to be translated

Calculations are the most laborious part, because the tools work differently. Qlik Sense loads through a script and links tables associatively via fields of the same name, while Power BI works best with a star schema. A straight copy therefore usually turns out worse than a rebuild with the same meaning.

Coming from Qlik, load scripts belong in data preparation, in Power Query or better in a data pipeline, and Set Analysis can often be expressed with CALCULATE and filters in DAX. Calculated fields and LOD expressions from Tableau become measures in the semantic model. A SAP BusinessObjects universe is essentially a semantic model, and Web Intelligence documents become reports.

Taking tables over one to one gets you little. We rebuild the data model and define the metrics centrally. That takes longer than a copy but makes every later report easier. Hidden calculations in scripts and reports often only surface during reconciliation, so we plan time for that.

AI assistants write usable drafts of DAX and explain existing formulas. The results often look right and still return different numbers, for instance because the filter context is different. We use them only as a starting point and test every formula against the legacy system’s values.

When an end of maintenance sets the pace

For SAP BusinessObjects, standard maintenance is ending for older releases. SAP continues the platform with newer releases, so replacing it is an option, not an obligation. Some older components are no longer part of the new releases, which can mean effort even for an upgrade. Standard maintenance for SAP BW 7.5 is also ending, with extended maintenance available at extra cost. The paths lead to BW/4HANA, to SAP Business Data Cloud or to a data warehouse outside SAP, among others.

For Power BI Premium, Microsoft is retiring capacities with P-SKUs and plans Fabric capacities (F-SKUs) as the successor. Existing contracts run to the end of their term, and per-user licenses such as Pro or Premium Per User are not affected. The changeover is comparatively small: you buy the new capacity first, assign the workspaces, and check refreshes, reports and gateways. Maintenance deadlines and contract terms are best clarified directly with the vendor. Moves between cloud data platforms are only touched on here.

Parallel running, training and shutdown

Old and new systems run side by side for a while. We reconcile the key metrics across several month and quarter closes, because some errors only show up at the annual close. Every deviation gets resolved: an error in the new model, an error in the old one, or a deliberately changed definition. What gets documented is which number applies.

At the same time the teams are trained, ideally with examples from their own reports, and contacts in the business teams help with the switch. Without an introduction, many go back to their old spreadsheets. Shutdown gets a fixed cutover date, otherwise you pay for both systems and nobody knows which one counts. After that the legacy system runs read-only and archived and is finally switched off, licenses and access included.