How would you design a Power BI data model for a new business domain?

Start by clarifying the business questions, the grain of each table, and how users will slice the data, then build a star schema with facts in the center and dimensions around it. Explain that you aim for one-to-many relationships from dimensions to facts, single-direction filter flow, and a clean Date dimension because time intelligence comes up in almost every report. For example, for sales reporting you might create a Sales fact table at the line-item level with keys like DateKey, ProductKey, CustomerKey, and measures like Sales Amount, Quantity, and Margin. Then you add dimension tables for Date, Product, Customer, Store, and Sales Rep, and you mark the Date table as a date table so YTD and YoY measures behave as expected. Call out what you avoid, such as many-to-many relationships without a strong reason, bi-directional filtering everywhere, and mixing multiple grains in one fact table. If the source data is messy, mention you will handle shaping in Power Query and keep the model simple so DAX stays readable and performance stays stable.

What is the difference between a calculated column and a measure in Power BI, and when would you use each?

A calculated column is computed during data refresh and stored in the model, while a measure is calculated at query time based on filter context. You use measures for aggregations that need to respond to slicers and visuals, and you use calculated columns when you need row-level attributes for relationships, grouping, sorting, or filtering. For example, “Sales Amount” should be a measure like SUM(Sales[Amount]) so it changes by date, region, and product. A “Customer Segment” derived from a rule might be a calculated column if you need it as a slicer or to create a relationship to a segment table. Mention tradeoffs, columns increase model size and can slow refresh, while measures can be harder for beginners because they rely on context. A helpful tip is to default to measures for numeric logic, then add columns only when you truly need a stored attribute for the model.

Explain row context vs filter context in DAX, and how context transition works.

Row context is “the current row” when DAX evaluates a row-by-row expression, which is common in calculated columns and iterator functions like SUMX. Filter context is the set of filters coming from visuals, slicers, page filters, and relationships, which is what makes measures respond to the report. A simple example is a measure like Total Sales = SUM(Sales[Amount]), where filter context from a Region slicer changes which rows are included. For row context, you might compute Line Margin = Sales[Revenue] - Sales[Cost] as a calculated column, or use SUMX(Sales, Sales[Revenue] - Sales[Cost]) when you need row-by-row math inside a measure. Context transition happens when CALCULATE turns row context into filter context, which is why measures inside iterators can behave differently once you wrap logic in CALCULATE. A common mistake is using EARLIER or complex patterns when a clearer approach is to use variables and explicit CALCULATE filters so you can explain the intent during the interview.

How do you write a Year-over-Year (YoY) and Year-to-Date (YTD) measure in DAX?

Start by confirming you have a proper Date table, it is marked as a date table, and it has a continuous range of dates. Then explain you will build a base measure first, and layer time intelligence on top so the logic stays modular. For example, you can create Total Sales = SUM(Sales[Amount]). Then YTD Sales = TOTALYTD([Total Sales], 'Date'[Date]) and YoY Sales = CALCULATE([Total Sales], SAMEPERIODLASTYEAR('Date'[Date])). If you need YoY % you can add YoY % = DIVIDE([Total Sales] - [YoY Sales], [YoY Sales]). Point out common pitfalls, such as using an auto date table unintentionally, missing dates in the Date table, or mixing fiscal calendars without defining them. If the business uses a fiscal year, mention you will adjust the Date table and use TOTALYTD with the year-end date parameter or build custom logic so the measures match finance expectations.

How would you optimize a slow Power BI report?

Start by narrowing down whether the bottleneck is data refresh, model size, or visual query performance, then work from the model outward. You can mention using Performance Analyzer to identify slow visuals, checking DAX Studio for expensive queries, and reviewing the data model for high cardinality columns and complex relationships. A practical example is removing unnecessary columns, changing text columns with many unique values to keys, and using a star schema so filters flow predictably. On the DAX side, you might replace complex iterators with simpler aggregations, reduce the use of DISTINCTCOUNT on huge tables, and use variables to avoid repeated calculations within a measure. Also call out report design choices, too many visuals on a page, high interaction counts, heavy custom visuals, and slicers on high-cardinality fields can all slow things down. If DirectQuery is involved, mention query folding, limiting visuals that generate many queries, and working with the source system to add indexes or summary tables.

What is Power Query, and how do you decide what logic belongs in Power Query vs DAX?

Power Query is for data ingestion and transformation during refresh, while DAX is for calculations at query time inside the model. A good rule is to do repeatable shaping steps in Power Query, then use DAX for business metrics that must react to filters and user selections. For example, removing nulls, splitting columns, parsing dates, standardizing product names, and joining reference tables are typically Power Query tasks. Measures like margin %, YoY growth, rolling 12 months, and dynamic segmentation are better in DAX because they depend on filter context. Mention you will keep transformations that reduce data volume early, because that improves refresh time and model size. Also note that some transformations should stay in the source system when possible, but in an interview you can say you choose the layer that keeps logic transparent and easier for the team to maintain.

Can you explain relationships in Power BI, including active vs inactive relationships and when to use USERELATIONSHIP?

Explain that relationships define how filters propagate between tables, and that Power BI allows one active relationship between two tables at a time. Active relationships drive filter flow automatically, while inactive relationships are ignored unless you activate them inside a measure. A common example is having a Sales table with both Order Date and Ship Date related to the Date dimension. You keep Order Date as the active relationship for most reporting, then create a measure like Ship Date Sales = CALCULATE([Total Sales], USERELATIONSHIP(Sales[ShipDate], 'Date'[Date])) when you need analysis by ship date. Call out that multiple date relationships are normal, but you should document which one is the default and create clear measure names so users do not mix concepts. Also mention you avoid bi-directional filters unless required, because they can create ambiguous filter paths and confusing results.

What are your go-to DAX patterns for handling blanks, divide-by-zero, and conditional logic?

Start by saying you aim for measures that return meaningful blanks instead of misleading zeros, and you handle errors explicitly so visuals do not show strange spikes. You can also mention you keep business rules readable by using variables and keeping measures small. For divide-by-zero, use DIVIDE(numerator, denominator) or DIVIDE(numerator, denominator, BLANK()) depending on what you want displayed. For conditional logic, use IF or SWITCH, for example, Margin % = DIVIDE([Total Margin], [Total Sales], BLANK()) and KPI Status = SWITCH(TRUE(), [Margin %] >= 0.3, "Green", [Margin %] >= 0.2, "Yellow", "Red"). A tip that interviewers like is describing how you test measures across edge cases, such as empty categories, new products with no history, or partial periods. Also mention formatting choices, like showing blanks as “N/A” in a tooltip rather than forcing zeros that can mislead business users.

How do you implement Row-Level Security (RLS) in Power BI?

Explain the approach, define roles, write DAX filters on dimension tables, test with “View as,” and then map users to roles in the Power BI Service. If the security rules are complex, mention using a security table that maps users to allowed regions, customers, or departments. For example, you might create a RegionSecurity table with UserEmail and RegionKey, relate it to the Region dimension, then define a role with a filter like RegionSecurity[UserEmail] = USERPRINCIPALNAME(). That way, a user only sees regions tied to their email, and the filter flows through the model to the fact tables. Call out common mistakes, such as applying RLS directly to a large fact table, which can hurt performance, or using usernames that do not match the identity in the service. You should also mention planning for admins and service accounts, and confirming how RLS should behave with drillthrough and export permissions.

How do you choose the right visuals and design an effective Power BI dashboard?

Start with the decision the dashboard should drive, then design around a few key questions and make the default view useful without extra clicks. Explain that you pick visuals based on the comparison type, trends over time, composition, distribution, and ranking, and you keep consistent formatting so users do not have to relearn the page. For example, you can use a line chart for trend, a bar chart for ranking products, a matrix for detail, and cards for a small set of headline KPIs. You can add drillthrough pages for deeper analysis, and use tooltips to provide context without cluttering the main canvas. Mention what you avoid, too many visuals on one page, heavy reliance on pie charts for many categories, and confusing color use that implies meaning where none exists. A practical tip is to validate with users by watching them complete a task, because real feedback usually catches confusing filters and unclear labels early.

Tell me about a time you had to fix a Power BI report that stakeholders did not trust.

Situation: A sales dashboard was showing totals that did not match the finance system, and stakeholders started exporting data and building their own spreadsheets. Task: I needed to find the root cause, align definitions, and rebuild confidence in the report. Action: I met with sales ops and finance to agree on a single definition for revenue, returns, and the date used for attribution. I audited the model, found that the report mixed order date and invoice date across visuals, then refactored measures and added a clear “As of” date plus a definitions page. Result: After validation against a set of sample invoices, the variance dropped to near zero for the agreed definition, and stakeholders stopped using side spreadsheets. Adoption improved, and the team used the report in weekly forecast meetings because the logic was transparent and documented.

Describe a time you improved Power BI report performance under a tight deadline.

Situation: A leadership report took 20 to 30 seconds to load during an executive meeting, and the team asked for fixes before the next meeting. Task: I had to cut load times quickly without changing the numbers or removing key insights. Action: I used Performance Analyzer to identify two visuals driving most of the delay, then reviewed their measures and underlying tables. I simplified the model to a clearer star schema, removed unused columns to reduce memory, and rewrote one measure to avoid expensive row-by-row iteration. Result: Page load time dropped to a few seconds in testing, and the report worked smoothly in the next meeting. The fixes also reduced refresh time, which gave the team more room to add new KPIs later without slowing everything down.

Tell me about a time you had to explain a complex DAX measure to a non-technical stakeholder.

Situation: A marketing lead questioned why a conversion rate changed when they filtered by channel, and the DAX included multiple steps and time logic. Task: I needed to explain the calculation in plain language so they could trust the number and use it correctly. Action: I broke the measure into smaller named measures, then walked through it using a simple example with one week and two channels. I explained that the denominator changed because the filter context limited which sessions were included, and I added a tooltip that listed the numerator and denominator values side by side. Result: The stakeholder agreed the logic matched the business definition, and they stopped treating the number as “buggy.” The tooltip and clearer measure names reduced repeat questions, and future changes were easier because the logic was modular.

power bi Interview Questions: Complete Guide

Power BI interviews usually test three things, your ability to model data well, write DAX that matches business logic, and build reports that people can actually use. You can expect a mix of discussion, whiteboard style questions, and a short hands-on exercise where you import data, create measures, and design visuals. If you feel rusty, that is normal. With a clear approach for modeling, DAX, and performance, you can give confident answers even when the question is open-ended.

Common Interview Questions

Behavioral Questions (STAR Method)

STAR Method: Structure your answers using Situation, Task, Action, and Result to tell compelling stories about your experience.

Questions to Ask the Interviewer

Show your interest by asking thoughtful questions

•What does success look like for this Power BI role after 30, 60, and 90 days?
•How is your semantic model managed today, and who owns definitions for key metrics like revenue and churn?
•What are the most common data quality issues your team faces, and how do you prefer to handle them, in the source, Power Query, or the model?
•Will I be working mostly in Import mode, DirectQuery, or a composite model, and what are your performance expectations for key reports?
•How do you handle governance, such as workspace structure, dataset certification, deployment pipelines, and documentation standards?

Interview Preparation Tips

1
Practice building a small star schema from a messy spreadsheet, then write 5 to 8 measures you can explain clearly, especially YTD, YoY, and rolling periods.
2
Bring a simple DAX explanation template, base measure first, then add CALCULATE filters, then confirm expected results with a tiny example table.
3
Before the interview, review one of your past reports and prepare a 2 minute story about the business goal, model choices, and one tradeoff you made.
4
If you get stuck, say what you would check next, such as the grain of the fact table, relationship direction, or whether the Date table is marked properly, then propose a test to validate it.

power bi Interview Questions: Complete Guide

Emily Thompson

Common Interview Questions

Behavioral Questions (STAR Method)

Questions to Ask the Interviewer

Interview Preparation Tips

Related guides

System Design Interview Questions in 2025

excel Interview Questions: Complete Guide

cloud computing Interview Questions: Complete Guide

Linux Interview Questions in 2025

power bi Interview Questions: Complete Guide

Emily Thompson

Common Interview Questions

Q1How would you design a Power BI data model for a new business domain?

Q2What is the difference between a calculated column and a measure in Power BI, and when would you use each?

Q3Explain row context vs filter context in DAX, and how context transition works.

Q4How do you write a Year-over-Year (YoY) and Year-to-Date (YTD) measure in DAX?

Q5How would you optimize a slow Power BI report?

Q6What is Power Query, and how do you decide what logic belongs in Power Query vs DAX?

Q7Can you explain relationships in Power BI, including active vs inactive relationships and when to use USERELATIONSHIP?

Q8What are your go-to DAX patterns for handling blanks, divide-by-zero, and conditional logic?

Q9How do you implement Row-Level Security (RLS) in Power BI?

Q10How do you choose the right visuals and design an effective Power BI dashboard?

Behavioral Questions (STAR Method)

B1Tell me about a time you had to fix a Power BI report that stakeholders did not trust.

B2Describe a time you improved Power BI report performance under a tight deadline.

B3Tell me about a time you had to explain a complex DAX measure to a non-technical stakeholder.

Questions to Ask the Interviewer

Interview Preparation Tips

Related guides

System Design Interview Questions in 2025

excel Interview Questions: Complete Guide

cloud computing Interview Questions: Complete Guide

Linux Interview Questions in 2025