Microsoft Power BI Data Analyst Associate Study Guide

Chapter 3. Modeling Data

Power BI is one of the most wonderfully flexible products ever created. It is simple and elegant by design, yet deep and complex in areas that defy explanation in simple terms. You can build a simple report very quickly.

Some seasoned professionals argue that Power BI is frequently used incorrectly. It’s true that many users create simple reports and visualizations using techniques that will not work at scale; when they need to scale up, they find they must start over and use completely different design techniques. This is avoidable. The purpose of this chapter is to put you on track to use Power BI correctly. Specifically, it will show you how to design data models in accordance with the way Power BI was designed to work at scale.

The Semantic Data Model

When the Power BI service was introduced, the priority was to streamline the experience for self-service citizen report developers: business users who create reports for themselves and their colleagues. From a technical perspective, creating a “report” in Power BI Desktop actually means that you are creating both a semantic data model and a report. Prior to 2024, published models in the service were called datasets, and now, appropriately, they’re called semantic models.

When you create a new report file (saved in .pbix or .pbip file format, which defines a Power BI Project) in Power BI Desktop, a semantic model is created. When you publish to the Power BI service, it deploys both a semantic model and ...