Power BI scenario-based interview questions, with answers

By ·

Illustrated BI developer at a laptop surrounded by paper cut-out charts, filters and a padlock

Power BI scenario based interview questions ask how you would handle real problems: a slow report, a total that does not match its rows, regional row-level security, fiscal year-to-date, mixed-grain budgets and failed refreshes. Strong answers follow one pattern: clarify the symptom, diagnose with a tool, fix the root cause, then explain how you prevent it happening again.

Power BI interviews have shifted. Fewer panels ask you to define a slicer, and more ask what you would do when a report is slow, a total looks wrong or a regional manager can see data they should not. These scenario questions test whether you have actually built and supported reports, not just followed a tutorial.

Below are 16 scenarios that come up for Power BI developer, BI analyst and data analyst roles, each with a short model answer. They are grouped by theme: performance, modelling, DAX behaviour, security and refresh. If you want definitions and pure DAX questions, start with our Power BI developer interview questions and come back here for the scenarios.

How do you answer a scenario-based Power BI question?

Interviewers are not looking for one magic feature. They want to hear a diagnosis order. A reliable pattern has four steps.

  1. Clarify the symptom: which report, which users, since when, and what changed recently.

  2. Diagnose with a tool, not a guess: Performance Analyzer, refresh history, DAX Studio or the model view.

  3. Fix the root cause, and say why that fix works.

  4. Prevent a repeat: a model review, a naming standard, monitoring or short documentation.

Keep each answer to about a minute, then offer to go deeper. That leaves room for the follow-up question, which is where most of the scoring happens.

What do you do when a Power BI report is slow?

1. A report page takes 20 seconds to load. Where do you start?

Open Performance Analyzer in Power BI Desktop, refresh the visuals and see which visual is slow and whether the time sits in the DAX query, the visual display or elsewhere. If DAX is the problem, copy the query into DAX Studio and check the server timings. Common fixes are fewer visuals per page, simpler measures that avoid iterating large tables, removing unused high-cardinality columns and moving to a star schema.

2. The PBIX file is huge and refresh takes too long. What do you change?

Check which columns use the most memory, for example with VertiPaq Analyzer in DAX Studio. Remove columns nobody reports on, split a datetime column into date and time or drop the time part if nobody needs it, set correct data types, and push calculated columns back to Power Query or the source. For very large fact tables, set up incremental refresh so only recent data reloads.

3. Would you use Import or DirectQuery for this report?

Import is usually fastest for users and supports all DAX, but data is only as fresh as the last refresh. DirectQuery leaves data in the source and suits near real-time needs or very large data, but every visual sends queries to the source, so speed depends on that system. A composite model lets you keep a big fact table in DirectQuery and small dimensions in Import. Ask how fresh the data must be and how large it is before choosing.

Which data modelling scenarios come up most?

4. Your model is one wide flat table from Excel. What would you do?

Reshape it into a star schema in Power Query: one fact table with numeric values and keys, plus dimension tables for date, product, customer and region. Star schemas compress better, keep DAX simpler and avoid ambiguous filter paths. Add a dedicated date table and mark it as a date table so time intelligence works reliably.

5. Budget is stored by month and category, but sales are daily by product. How do you compare them?

The two fact tables have different grains, so do not relate them to each other. Relate both to shared dimensions at the level they have in common: the Date table, with budget linked through a month start date, and a Category dimension. Visuals by month and category then show both measures correctly. If a user drills down to product, the budget measure should return blank or use an allocation rule the business has agreed.

6. Power BI says a relationship cannot be one-to-many because of duplicate keys. Why?

The key column in the dimension has duplicates. Look for true duplicates, blanks and values that differ only by trailing spaces or case. Power Query is case-sensitive but the data model is not, so ABC and abc both survive Remove Duplicates in Power Query and then collide in the model. Trim, clean and standardise case in Power Query, then remove duplicates.

7. A slicer is not filtering one of the visuals. What do you check?

Confirm the relationship exists, is active and has the right cross-filter direction, and that the slicer uses a column from the dimension, not from a different fact table. Then check Edit interactions on that page, and whether the measure uses ALL or REMOVEFILTERS in a way that ignores the slicer.

Why is this DAX number wrong? Common calculation scenarios

8. The total row shows a number that is not the sum of the rows above it.

This is the most popular DAX scenario. Say sales staff earn a 5 percent bonus only when their sales cross 1,00,000. The first version of the measure is Bonus = IF([Sales] > 100000, [Sales] * 0.05, 0). Here is what the table visual shows.

Salesperson

Sales

Bonus (first version)

Correct bonus

Priya

80,000

0

0

Arjun

1,50,000

7,500

7,500

Sneha

1,20,000

6,000

6,000

Total

3,50,000

17,500

13,500

Measures are evaluated in the filter context of each cell, including the total. At the total row, [Sales] is 3,50,000, which is above the threshold, so the measure returns 5 percent of the grand total instead of adding up the individual bonuses. The fix is to iterate over salespeople.

Bonus Fixed = SUMX(VALUES(Salesperson[Name]), IF([Sales] > 100000, [Sales] * 0.05, 0))

SUMX walks through each salesperson, and because [Sales] is a measure, context transition evaluates it for that one person. The total now adds the individual bonuses and shows 13,500.

9. Profit margin by region looks wrong when you average it.

Averaging row-level margins weights every transaction equally, so a tiny order counts as much as a huge one. Margin should be a ratio of totals: Margin % = DIVIDE(SUM(Sales[Profit]), SUM(Sales[Revenue])). DIVIDE also returns blank instead of an error when revenue is zero.

10. Finance wants year-to-date on the Indian fiscal year, April to March.

Use TOTALYTD with a year end date: Sales FYTD = TOTALYTD([Total Sales], 'Date'[Date], "3/31"). It needs a proper date table with continuous dates, marked as a date table. Also add a fiscal year column to the date table, so visuals group by fiscal year rather than calendar year.

11. A manager wants this month compared with the same month last year.

Write Sales LY = CALCULATE([Total Sales], SAMEPERIODLASTYEAR('Date'[Date])), then a growth measure with DIVIDE([Total Sales] - [Sales LY], [Sales LY]). Point out that the current month is partial, so comparing half a month with a full month misleads. Either compare up to the same day or show completed months only.

12. Users want one chart that switches between revenue, units and margin.

Use field parameters, which let users pick the measure from a slicer without extra DAX. The older pattern is a disconnected table with a SWITCH measure based on SELECTEDVALUE. Both are acceptable answers; say which one you have used in a real report.

How do you handle row-level security scenarios?

13. Each regional manager should see only their own region.

Use row-level security. For a few fixed roles, static rules on the Region table are enough. For many users, use dynamic RLS: a mapping table of user emails and regions, and a rule on the Region table such as Region[Region] IN SELECTCOLUMNS(FILTER(UserRegion, UserRegion[Email] = USERPRINCIPALNAME()), "R", UserRegion[Region]). Hide the mapping table, and test with View as role in Desktop and in the service before release.

14. A manager says RLS is broken because they can see every region.

Check their workspace role first. RLS applies to users with Viewer permission. Workspace Admins, Members and Contributors are not restricted by it. Share the content through an app or give them Viewer access, then test again with their account.

What if a scheduled refresh fails?

15. Last night's refresh failed. What do you do?

Open the refresh history in the service and read the actual error. Common causes are expired data source credentials, an on-premises data gateway that is offline or out of date, a renamed column or table at the source, and timeouts on large tables. Fix the cause, run a manual refresh, and make sure failure notifications reach the owner before the business notices.

16. A fact table holds five years of data, but only recent rows change.

Set up incremental refresh. Create RangeStart and RangeEnd parameters of type Date/Time in Power Query, filter the table on them, then define a policy such as store five years and refresh the last ten days. The policy takes effect after you publish to the service. It works best when the filter folds back to the source, so the database returns only the rows needed.

How do you say it in the interview?

Here is a full answer to the slow report question using clarify, diagnose, fix and prevent. The numbers are the candidate's own example, which is what makes it believable.

First I would confirm which page is slow and for whom, because a page that is slow only for some users can point to RLS or the gateway. Then I would run Performance Analyzer to find the slowest visual and check whether the time is in the DAX query. In my last project for a retail client, one page had 18 visuals and a measure that used FILTER over the whole sales table. I replaced it with a simple CALCULATE filter on a dimension column, cut the page to 9 visuals with a drill-through for detail, and load time dropped from about 15 seconds to under 4. After that we added a short model review checklist before any report went live.

Practise these out loud, not just in your head. A few timed mock interviews will show you which answers run long. Our technical interview tips cover how to handle a scenario you have not seen before. If your Power BI round happens on a video call, the Interview Copilot for technical interviews drafts structured answers from your own resume; always follow each employer's interview rules.

Already working in a Power BI role and stuck on a real model? Power BI job support combines AI with expert help on live tickets. To rehearse these 16 scenarios with feedback on every answer, try InterviewPlan AI free with 5 live minutes and 100 AI credits, no card needed.

Frequently asked questions

What are scenario-based questions in a Power BI interview?

They describe a real situation, such as a slow report, a wrong total or a failed refresh, and ask what you would do. The interviewer is checking your diagnosis order and practical experience, not just definitions. Answer by clarifying the symptom, naming the tool you would use, fixing the root cause and explaining how you would prevent it.

Which Power BI scenario question is asked most often?

Report performance and totals that do not match their rows come up most. Be ready to explain Performance Analyzer, star schemas and filter context, and to fix a total with SUMX over a dimension. Row-level security and fiscal year-to-date are close behind, especially for roles supporting Indian finance teams.

Are Power BI scenario questions asked for freshers?

Freshers usually get simpler versions, such as choosing a chart, cleaning data in Power Query or writing a basic measure. Even so, one modelling or DAX scenario is common. If you have a college or personal project, explain a real problem you hit in it and how you solved it, using your own numbers.

Do Power BI interviews include a live task?

Many do. You may get a sample dataset and 30 to 60 minutes to build a small report, or be asked to share your screen and fix a model. Narrate your thinking as you work, keep the model clean with a date table and star schema, and explain any trade-offs you make because of time.

How should I prepare for Power BI scenario questions in a week?

Rebuild one report end to end with a proper star schema and date table, then deliberately break it: add a bad measure, a duplicate key and an RLS rule. Fixing your own mistakes teaches the scenarios faster than reading. Finish by rehearsing answers aloud in timed mock interviews.

P.S.

Reading helps. Practising out loud helps more.

Start free with 5 live minutes and 100 AI credits for Resume AI, Job Search, Auto-Apply, mock interviews and Behavioral Coach.