Power BI Interview Questions

Analogy categoryAnalogies

Fundamentals Questions

Interviewers typically open with fundamentals to confirm you understand the platform's building blocks: the difference between Power BI Desktop, Service, and Mobile; the three core components (Power Query for ETL, the Tabular/VertiPaq data model, and DAX for calculations); and the distinction between Import, DirectQuery, and Composite storage modes. A strong candidate doesn't just define these terms but explains trade-offs — for example, that Import mode loads a compressed in-memory copy refreshed on a schedule, offering the best performance but with data-freshness limits, whereas DirectQuery queries the source live on every interaction, trading performance for real-time accuracy.

🏏

Cricket analogy: Explaining Import versus DirectQuery is like explaining the difference between a pre-recorded match highlights reel (fast, but not live) and a live broadcast feed (real-time, but dependent on the live signal quality).

DAX and Data Modeling Questions

A near-universal interview question is 'explain the difference between a calculated column and a measure, and when you'd use each' — the correct answer covers row-context evaluation and storage for calculated columns versus filter-context, on-demand evaluation for measures. Another common one is 'what does CALCULATE do and why is it central to DAX?' — the answer should explain that CALCULATE evaluates an expression in a modified filter context, making it the mechanism behind time intelligence, ALL/ALLEXCEPT overrides, and most non-trivial business logic. Expect a live DAX exercise too, such as writing a measure for running total, percent of total, or year-over-year growth on the spot.

🏏

Cricket analogy: CALCULATE modifying filter context is like a third umpire overturning an on-field decision after reviewing specific replay angles, changing the outcome based on a deliberately narrowed context.

DAX

-- A common live-coding interview question: percent of total
Product % of Total Sales =
DIVIDE (
    [Total Sales],
    CALCULATE ( [Total Sales], ALL ( Product ) )
)

-- Running total by date
Running Total Sales =
CALCULATE (
    [Total Sales],
    FILTER ( ALLSELECTED ( 'Date' ), 'Date'[Date] <= MAX ( 'Date'[Date] ) )
)

Performance and Architecture Questions

Senior-level interviews probe how you'd diagnose and fix a slow report: expect questions about Performance Analyzer, identifying whether the bottleneck is the DAX query engine or the visual rendering, and understanding when to use aggregation tables or Composite models to blend a small Import-mode summary table with a large DirectQuery fact table. You should also be ready to discuss incremental refresh (partitioning a large fact table by date so only recent partitions reprocess on refresh) and the role of Premium/Fabric capacity in supporting larger models, paginated reports, and dataflows shared across workspaces.

🏏

Cricket analogy: Diagnosing whether a report's slowness is the DAX engine or visual rendering is like a coach diagnosing whether a batter's slow strike rate comes from footwork or shot selection — different symptoms need different fixes.

A strong interview answer on performance always names the actual tool: Performance Analyzer (Desktop, per-visual query timing) and DAX Studio (server timings, query plans) rather than describing optimization only in the abstract.

Scenario-Based Questions

Scenario questions test judgment under ambiguity: 'A stakeholder wants a report refreshed every 5 minutes on a 200-million-row table — how do you approach it?' expects you to weigh incremental refresh, DirectQuery/Composite options, and push back on unrealistic freshness requirements with a cost-benefit conversation. 'Two teams define "active customer" differently — how do you resolve it in the model?' expects an answer about a shared semantic model with clearly documented, business-approved measure definitions rather than letting each team build conflicting local calculations. These questions aren't testing DAX syntax; they're testing whether you can translate ambiguous business requirements into a defensible technical design.

🏏

Cricket analogy: Resolving two teams' conflicting 'active customer' definitions is like a board settling a dispute between two states' associations over qualification rules by publishing one official governing definition everyone must follow.

Avoid answering scenario questions with pure technical jargon and no business framing. Interviewers are listening for whether you'd ask clarifying questions ('what does the business actually need refreshed that fast, and why?') before jumping to a technical solution.

Know the difference between Power BI Desktop, Service, Mobile, and the roles of Power Query, the Tabular model, and DAX.
Be fluent explaining Import, DirectQuery, and Composite storage mode trade-offs.
Master the calculated column versus measure distinction and be ready to write DAX live.
Explain CALCULATE's role in modifying filter context as the backbone of most non-trivial DAX.
Be ready to discuss Performance Analyzer, DAX Studio, aggregation tables, and incremental refresh.
For scenario questions, show business judgment, not just syntax — ask clarifying questions first.
Advocate for a single shared semantic model to resolve conflicting metric definitions across teams.