Question 1
Difficulty: easy
Can you walk me through how you would build a new Power BI dashboard from a business requirement document?
Sample answer
I usually start by clarifying the business goal, the audience, and the decisions the report needs to support. Then I map the required KPIs to the source systems and identify any data quality or refresh constraints early. After that, I design the data model first, because a solid star schema makes everything easier downstream. I would build and validate the dataset in Power Query, define measures in DAX, and then create a layout that highlights the most important insights before adding decorative elements. I also like to review a first draft with stakeholders quickly so I can confirm the visuals, filters, and drill paths match their expectations. Before handoff, I test performance, security rules, and refresh behavior. My goal is always to deliver something that is accurate, fast, and usable, not just visually polished.
Question 2
Difficulty: medium
How do you decide whether to use a calculated column, a measure, or a Power Query transformation in Power BI?
Sample answer
I make that decision based on when the calculation needs to happen and how flexible it needs to be. If the logic is about shaping or cleaning the data before it lands in the model, I prefer Power Query because it keeps the model lean and improves maintainability. If the result should respond dynamically to filters, slicers, or user context, I use a measure in DAX. Calculated columns are useful when I need a value stored at row level for sorting, relationships, or fixed categorization, but I avoid overusing them because they increase model size. In practice, I try to push simple transformations upstream or into Power Query, reserve measures for analytics, and use calculated columns only when they solve a clear modeling need. That approach usually gives better performance and makes the report easier to support later.
Question 3
Difficulty: medium
Tell me about a time you had to improve the performance of a slow Power BI report.
Sample answer
In one project, users complained that a sales report took too long to open and the slicers were lagging. I started by checking the model size, number of columns, and relationships, then reviewed the DAX measures for anything overly complex. I found a few issues: unnecessary high-cardinality columns, some measures doing repeated calculations, and a model structure that was closer to a flat table than a proper star schema. I reduced the dataset to only needed fields, rewrote several measures using variables and simpler filter logic, and separated the fact and dimension tables more cleanly. I also reviewed Power Query steps to make sure folding was preserved where possible. After those changes, load times improved significantly and the report became much more responsive. The biggest lesson was that performance issues usually come from several small design decisions, not one single problem.
Question 4
Difficulty: easy
How do you handle a situation where business users want too many visuals on one report page?
Sample answer
I try to redirect the conversation from adding visuals to answering questions. If a page is overcrowded, users usually lose the main message and spend more time searching than analyzing. I would first ask what decisions the page should support and which metrics are truly essential. Then I would group related measures, use drill-through or bookmarks where needed, and split the story across multiple pages if that makes the experience clearer. I also like to use tooltips, summary cards, and well-designed filters to keep the page clean without removing useful detail. When I present that approach, I explain that a simpler page often creates better adoption because it reduces confusion and speeds up decision-making. In my experience, stakeholders are usually open to it once they see that the redesigned page actually makes the report more valuable.
Question 5
Difficulty: hard
How would you implement row-level security in a Power BI solution for different regional teams?
Sample answer
I would start by confirming the access rules with the business, because security should reflect how the organization actually operates. If the requirement is regional visibility, I typically create a security mapping table that links users or roles to regions. Then I apply row-level security either directly on the region dimension or through a bridge table if the setup is more complex. I prefer using dynamic security with USERPRINCIPALNAME when the environment needs to scale, because it avoids maintaining too many static roles. After building the role, I test it with multiple user scenarios in Desktop and then validate it again in the Service with test accounts. I also check whether the security design works with related reports or apps so there are no unexpected gaps. For me, the key is balancing ease of maintenance, correct data access, and clear documentation for support.
Question 6
Difficulty: medium
Describe your process for writing DAX when a simple aggregation is not enough.
Sample answer
When an aggregation gets more complex, I break the problem into smaller business rules before I write any code. I first define the filter context I need, the grain of the calculation, and whether the result should react to slicers or ignore some of them. Then I build the logic step by step, usually testing each part in a card or table visual before combining it into the final measure. I rely heavily on variables because they make DAX easier to read and often improve performance. If the calculation involves time intelligence, ranking, or conditional logic, I verify the result against a known sample set so I can be confident it matches the business rule. I also keep an eye on readability, because someone else may need to maintain the measure later. Good DAX is not just correct; it is also understandable and predictable.
Question 7
Difficulty: medium
How do you ensure data quality when building a Power BI solution from multiple source systems?
Sample answer
I treat data quality as part of the solution, not a separate cleanup task. At the beginning, I compare key fields across systems to identify mismatches in naming, dates, codes, and grain. In Power Query, I standardize formats, remove duplicates where appropriate, and flag suspicious records instead of hiding them. I also add validation checks for things like null values in required fields, invalid relationships, or totals that don’t reconcile between systems. When possible, I build a basic reconciliation page so business users can see source-to-report comparisons and trust the numbers. If I find recurring issues, I document them and work with the source system owners to improve the upstream process. I think the best Power BI developers are proactive about data quality because a beautiful report is still useless if the numbers cannot be trusted.
Question 8
Difficulty: easy
What would you do if a stakeholder insists that a number in the report is wrong, but your validation shows it is correct?
Sample answer
I would approach it calmly and treat it as a collaboration issue rather than a disagreement. First, I would ask the stakeholder to show me how they arrived at their number so I can compare their logic to the report logic. Often the mismatch comes from a different filter, date definition, or business rule rather than an actual error. I would then walk through the calculation with them using a small set of records so we can isolate where the difference appears. If the report is correct but the business expectation is different, I would update the requirements or document the agreed rule clearly. If I do find a real issue, I fix it quickly and share what changed. My aim is to make the conversation transparent and factual so the stakeholder feels heard and the final solution is trusted.
Question 9
Difficulty: easy
How do you decide which visuals and layout to use for an executive-level Power BI report?
Sample answer
For executive reporting, I focus on clarity, speed, and trend visibility. I usually start with a small set of headline KPIs at the top, then add trend charts, variance views, and a few supporting breakdowns that explain what is driving the numbers. I avoid clutter, heavy grids, and too many colors because executives usually need the main story quickly. I also think carefully about hierarchy: the most important insight should be visible without scrolling or hunting. If a chart is intended to show change over time, I choose visuals that make trends obvious, such as line or column charts, and I keep labels simple. I also make sure the report works well on different screen sizes if it will be viewed during meetings. A strong executive dashboard should answer the main questions in seconds and still allow deeper exploration when needed.
Question 10
Difficulty: medium
How do you collaborate with business analysts, data engineers, and end users during a Power BI project?
Sample answer
I try to keep collaboration structured but flexible. With business analysts and end users, I focus on clarifying requirements, definitions, and the decisions the report should support. With data engineers, I align on source availability, refresh timing, schema design, and any transformation that is better handled upstream. I like to stay involved early so I can spot issues before they become redesign work later. Throughout the project, I use short review cycles with real sample data, because that helps stakeholders react to something tangible instead of abstract mockups. I also document assumptions and open questions so everyone has a shared reference point. When communication is strong, the final report usually lands much closer to what the business actually needs. In my experience, the best Power BI work comes from treating the project as a team effort, not just a report-building task.
