Dataflows seemed like a great idea at the time. A no-code ETL layer, right inside Power BI, manageable by the BI team without involving data engineering. What could go wrong?
Quite a lot, as it turns out.
In this session, I'll share the real story of decommissioning Power BI Dataflows Gen1 across an enterprise environment and migrating to a proper data engineering layer built on Databricks. We'll cover why dataflows fail at scale — refresh cascades, maintenance overhead, hidden dependencies, and the moment you realize your "quick win" has become critical infrastructure nobody wants to own.
More importantly, we'll talk about the human side: how to communicate a deprecation to teams who built their reports on top of your dataflows, how to manage the migration without breaking production, and how to turn a painful decommissioning into an opportunity to modernize your entire data stack.
Key takeaways:

Why Power BI Dataflows struggle at enterprise scale and when to walk away
Migration patterns from Dataflows to Databricks ETL
Communication and change management for platform deprecations
How to design a data engineering layer that actually scales

Real ongoing decommissioning in a regulated financial services environment. Names anonymized. Some dataflows are still running.
| Target audience: BI developers, data engineers, Power BI admins, architects
| Preferred duration: 60 min
| Basic familiarity with Power BI & Dataflows recommended, however not required