Power BI Developer Interview Questions and Answers

Questions and model answers

9 questions you are likely to hear.

1. 1

   What is the difference between a calculated column and a measure?

   Why they ask: This is a foundation check: the wrong choice causes bloated models and wrong numbers.

   Sample answer: A calculated column is computed row by row at refresh time and stored in the model, so it increases model size and is evaluated in row context. A measure is computed at query time in the current filter context and is not stored. I use calculated columns when I need to slice or filter by the value, like an age band, and measures for aggregations such as total sales or margin percent. In one model I replaced eight calculated columns with measures and a small dimension table, which reduced the file size by roughly a third and improved refresh time.

2. 2

   Explain filter context and how CALCULATE changes it.

   Why they ask: Understanding context is the single biggest predictor of DAX skill.

   Sample answer: Filter context is the set of filters applied to the model when a measure is evaluated, coming from slicers, visual rows and columns, and report filters. CALCULATE evaluates an expression after modifying that context: it can add filters, replace them or remove them. For example, CALCULATE([Total Sales], Product[Category] = "Bikes") overrides any category filter on the visual, while CALCULATE([Total Sales], REMOVEFILTERS(Product)) gives the grand total for a percent of total measure. CALCULATE also performs context transition, turning row context into filter context, which is why it behaves differently inside iterators.

3. 3

   Why is a star schema recommended for Power BI?

   Why they ask: The panel checks modelling fundamentals and whether you can explain performance and correctness benefits.

   Sample answer: A star schema separates facts, such as sales transactions, from dimensions like date, product and customer, connected by one-to-many relationships. VertiPaq compresses dimension columns well, filters flow in one predictable direction and DAX stays simple. A single flat table repeats text values and makes time intelligence awkward, while snowflaked or many-to-many models cause ambiguous filter paths. On a retail project I converted a 60-column flat extract into a fact table and five dimensions, which made measures much easier to write and cut visual load times from around 8 seconds to under 2.

4. 4

   A report is slow. How do you troubleshoot it?

   Why they ask: Performance tuning is a daily task, so interviewers want a systematic approach.

   Sample answer: I start with Performance Analyzer to see which visuals are slow and whether time goes to the DAX query or rendering. I copy the slow query into DAX Studio, check server timings and look for heavy formula engine work, often from FILTER over a whole table or nested iterators. Then I check the model with VertiPaq Analyzer for high-cardinality columns such as datetime stamps or GUIDs. Typical fixes are splitting date and time, removing unused columns, replacing FILTER with simple predicates and reducing visuals per page. In one case these steps took a page from 15 seconds to about 3.

5. 5

   How do you implement dynamic row-level security?

   Why they ask: Security questions test both technical setup and awareness of data governance.

   Sample answer: I create a security table mapping user email to the regions or entities they can see, relate it to the relevant dimension, and define a role with a DAX filter such as Security[Email] = USERPRINCIPALNAME(). Filters then flow from the security table to the dimension and fact. I test with View as role in Desktop and in the Service with real test accounts before publishing. For a sales organisation with around 300 users, this replaced 12 separate regional reports with one, and access changes became a row update in the security table instead of a new report.

6. 6

   When would you use DirectQuery instead of Import mode?

   Why they ask: This checks architectural judgement and understanding of trade-offs.

   Sample answer: Import is my default because it is fastest and supports full DAX. I consider DirectQuery when data is too large to import, when users need near real-time data, or when security rules require data to stay in the source. The trade-offs are slower visuals, dependence on source performance and some DAX limitations. Often a composite model is best: for a logistics client I kept historical shipments in Import with incremental refresh and used DirectQuery only for today's shipment table, so operations saw live status while trend pages stayed fast.

7. 7

   What is query folding and why does it matter?

   Why they ask: Interviewers check that you understand Power Query performance, not just its interface.

   Sample answer: Query folding means Power Query translates transformation steps into a native query, like SQL, so the source database does the work instead of the Power BI engine. It matters because folded queries refresh much faster and are required for incremental refresh. Steps like filtering, removing columns and grouping usually fold against SQL Server, while adding an index column or some custom M functions break folding. I check with View Native Query and keep folding steps first. Reordering steps in one dataflow brought refresh time down from about 40 minutes to 6.

8. 8

   Write a measure for year-over-year sales growth.

   Why they ask: Time intelligence is asked in almost every Power BI interview.

   Sample answer: I would define a variable for current sales as [Total Sales] and prior year sales as CALCULATE([Total Sales], SAMEPERIODLASTYEAR('Date'[Date])), then return DIVIDE(current minus prior, prior). DIVIDE handles division by zero by returning blank. This needs a continuous date table marked as a date table and related to the fact. I would also handle partial periods: if the current month is incomplete, I either compare to the same number of days last year or show growth only for completed months, and I would confirm with the business which they prefer.

9. 9

   Tell me about a Power BI solution you delivered end to end.

   Why they ask: This shows ownership, stakeholder handling and measurable value.

   Sample answer: For a manufacturing client, finance spent two days each month compiling plant performance in Excel. I gathered requirements with the controllers, connected to SQL Server and SAP extracts through a gateway, and built a star model with a date, plant and product dimension. I wrote around 40 measures, including variance to budget and rolling 12-month cost per unit, and set up row-level security per plant. After UAT we published it as an app with scheduled refresh. Monthly reporting dropped to a few hours, and plant managers started using the drill-through page in their weekly reviews.

Keep going.

• Tableau Developer Interview Questions and Answers
• Data Analyst Interview Questions and Answers
• SQL Developer Interview Questions and Answers
• Power BI Job Support Online for BI Developers
• Technical Interview Tips That Actually Help
• AI Copilot for Technical Interviews