What is Power BI? How it should read data
Power BI components, Import versus DirectQuery, where the data model should come from, and why connecting it straight to your ERP tends to backfire.
Power BI is Microsoft's business intelligence and visualisation product. It has three parts: Power BI Desktop for modelling, the Power BI Service for publishing and sharing, and mobile apps for consumption.
Where the data should come from
Power BI connects to hundreds of sources directly — and that convenience creates the most common mistake. If every report opens its own query against the ERP, three problems follow: the live system slows down, each author writes their own version of a metric, and no one keeps history.
The healthier pattern: move the data once, apply business rules in the warehouse, and let Power BI read consistent tables. 'Net sales' is then defined in exactly one place.
Import or DirectQuery?
- Import: data is loaded into Power BI's own engine. Fast to use, needs a refresh schedule.
- DirectQuery: every interaction queries the source. Data is current, but source performance is felt by every user.
- Composite: aggregates imported, detail on demand.
What a warehouse changes
- Report speed stops depending on the load of the operational system
- The same metric returns the same number in every report
- Past periods stay reproducible even when the source is corrected
- Data from several systems joins in one model
Floemdata produces exactly this layer — Power BI, Tableau, Qlik or Excel read the same reconciled tables through one API.