1.1. From Raw Data to Decisions: The Power BI Pipeline
💡 First Principle: Raw data answers no questions on its own. Before anyone can make a decision from it, data must be gathered, cleaned, related, calculated, displayed, and delivered to the right people. Power BI is a pipeline that performs those steps in a fixed order, and each exam domain is one segment of that pipeline. When you can place a requirement on the pipeline, you know which tool solves it.
Think about what happens when a sales director asks, "Are we beating last year in the Midwest?" Somebody has to pull orders from the ERP system and targets from a spreadsheet (connect), fix the region names that were typed three different ways (transform), link orders to dates and regions so "last year" and "Midwest" mean something (model), compute the comparison (calculate), draw it so the answer is obvious in two seconds (visualize), and get it in front of the director without exposing other regions' numbers (share and secure). Skip a step and the answer is either missing or wrong.
The color bands map directly to the exam: blue is Prepare the data, purple is Model the data, teal is Visualize and analyze, and yellow is Manage and secure. A useful exam habit is to ask, for every stem, "which segment is broken or being asked about?" A question about inconsistent region names belongs to transform, not DAX. A question about users seeing other regions belongs to security, not filters on the report page.
⚠️ Common Misconception: A Power BI report and a semantic model are the same thing. They are separate layers. The semantic model holds data, relationships, and calculations; the report is a visual layer that sends queries to a model. One model can feed many reports, and publishing a single .pbix file creates two distinct items in the service.
