Power BI Last-Minute Revision

30 quick revision notes for Power BI

1. Power BI Desktop is for building reports, the Service for publishing and sharing, and the gateway for refreshing on-premises sources.
2. A report is multi-page and interactive on one semantic model; a dashboard is a single page of pinned tiles from many reports.
3. Import mode is fastest but needs refresh, DirectQuery queries the source live, and Direct Lake reads OneLake Delta tables directly.
4. Datasets are now called semantic models; one shared model can serve many thin reports with consistent measures.
5. Query folding pushes Power Query steps to the source as native queries and is required for efficient incremental refresh.
6. Merge queries join tables side by side like SQL joins; append stacks rows of similar tables like UNION ALL.
7. Use Unpivot Other Columns to turn month columns into rows so new months are handled automatically.
8. A star schema has fact tables at a clear grain surrounded by dimension tables related one-to-many in a single direction.
9. Avoid bi-directional relationships by default; use CROSSFILTER inside a specific measure when two-way filtering is needed.
10. Only one relationship between two tables can be active; use USERELATIONSHIP in CALCULATE to activate an inactive one.
11. Mark a contiguous date table as a date table and turn off Auto date/time to keep the model small.
12. Measures are calculated at query time in filter context; calculated columns are computed at refresh and stored in memory.
13. Row context iterates rows; filter context comes from slicers, visuals and CALCULATE; context transition converts row context into filters.
14. CALCULATE changes filter context, and boolean filter arguments replace existing filters on the same column unless KEEPFILTERS is used.
15. SUM aggregates one column while SUMX evaluates an expression row by row and then sums the results.
16. Use DIVIDE instead of the slash operator to handle zero or blank denominators safely.
17. Variables in DAX are evaluated once where defined, which improves readability and performance of repeated expressions.
18. Totals in a visual are evaluated in their own filter context, so non-additive measures may not equal the sum of rows.
19. TOTALYTD accepts a year-end date such as "31/3" for the Indian April to March fiscal year.
20. Calculation groups apply reusable logic like YTD and PY to any measure through SELECTEDMEASURE, reducing measure count.
21. Field parameters let users switch the measure or dimension shown in a visual with a slicer.
22. Drill-down moves through a hierarchy in one visual; drill-through opens a separate filtered detail page.
23. Row-level security filters rows by role; dynamic RLS uses USERPRINCIPALNAME with a user-to-region mapping table.
24. RLS applies only to viewers; workspace Admins, Members and Contributors can see all data in the model.
25. Incremental refresh uses RangeStart and RangeEnd parameters to refresh only recent partitions of large tables.
26. Pro allows up to 8 scheduled refreshes per day, while Premium and Fabric capacity allow up to 48.
27. Apps with audiences are the recommended way to distribute reports to large groups of business users.
28. VertiPaq compresses columns by dictionary and run-length encoding, so high-cardinality columns use the most memory.
29. Use Performance Analyzer to find slow visuals and DAX Studio server timings to analyse their queries.
30. Aggregation tables in Import mode with Dual dimensions speed up DirectQuery models on very large data.

Night-before checklist

• Revise your saved questions on this page once, aloud.
• Read the quick notes and mark any you cannot explain in one line.
• Rehearse a two-minute self-introduction and two project stories.
• Keep CV, ID, laptop charger and interview link ready; sleep on time.