Power BI Interview Questions for 3 Years Experience
40 questions with answers for candidates with 3 years of experience — 4 easy, 31 medium, 5 hard.
What interviewers expect: 3 years experience
Three-year candidates should own complete solutions. Questions move to incremental refresh, DAX Studio and VertiPaq Analyzer, calculation groups, deployment pipelines, composite models and handling large data. Expect scenario rounds on slow reports, refresh failures and conflicting numbers between departments, where your structured troubleshooting approach matters most. Some companies add a take-home case study on a sample dataset.
This set is weighted towards Power BI Service, Sharing and Security (8), Performance Optimisation and Administration (7), Data Modelling (6), Advanced DAX and Time Intelligence (5).
Incremental refresh DAX Studio tuning Deployment pipelines Troubleshooting scenarios
Solved at this levelStart practicing
3 QUESTIONS FREESign up to unlock the full question bank and track your progress.Unlock every question and answer, then continue practising company-wise and package-wise.
Parameters are named values used inside queries, such as server name, database, file path or a date range. They let you switch from a development to a production database without editing each query, and they are required for incremental refresh through RangeStart and RangeEnd. Parameters can be changed in the Service under semantic model settings, which helps deployment pipelines between environments.
02
What is the difference between static and dynamic RLS?
Static RLS hard-codes values in each role, such as one role per region, which becomes hard to manage as regions grow. Dynamic RLS uses one role with a filter based on the logged-in user, such as Users[Email] = USERPRINCIPALNAME(), with a security mapping table linking users to the regions or managers they may see. Dynamic RLS scales better and is driven by data rather than role edits.
03
A report page takes 25 seconds to load. Walk me through how you would fix it.
I would run Performance Analyzer to find the slowest visuals, copy their DAX queries into DAX Studio and check server timings. Common fixes are rewriting measures that iterate large fact tables, removing bi-directional relationships, reducing visuals on the page, and shrinking the model by removing high-cardinality columns. If the source is DirectQuery, I would add aggregations or move to Import. Then re-measure after each change.
04
Split a datetime column to reduce cardinality in Power Query.
= Table.AddColumn(Source, "OrderDateOnly", each DateTime.Date([OrderDateTime]), type date)
= Table.AddColumn(Prev, "OrderHour", each Time.Hour([OrderDateTime]), Int64.Type)
Then remove the original OrderDateTime column. A datetime with seconds can have millions of unique values, while date and hour columns have only thousands and 24 values, compressing much better.
05
A YTD measure returns blank for future months on a chart. Users want the line to stop at today. How?
The issue is the opposite for many teams: YTD repeats the last value for future months. I would add a condition so the measure returns BLANK when the period starts after the last date with data, for example IF(MIN('Date'[Date]) > [Last Sales Date], BLANK(), TOTALYTD(...)). Alternatively add a column flagging dates up to today in the date table and filter the visual on it.
06
Tell me about a time you found a wrong number in a published report.
Situation: a regional head noticed our margin report did not match finance. Task: find and fix it quickly. Action: I compared numbers at product level, found returns were double-subtracted because a measure used both a net amount column and a returns table. I fixed the measure, added a reconciliation page against finance totals and informed users. Result: numbers matched to the rupee and the reconciliation check became standard.
In Desktop use Modeling, View as, select a role and optionally Other user with an email to simulate dynamic RLS, and check visuals show only expected rows. In the Service open the semantic model's Security settings, select a role and use Test as role. Test with real sample users from different regions and a user with no mapping, who should see no data, and confirm totals match expectations.
Row context means DAX is evaluating one row at a time, which happens in calculated columns and iterators like SUMX, so you can refer to column values of the current row. Filter context is the set of filters applied to the model from slicers, visual rows, columns and CALCULATE. Measures are evaluated in filter context. Row context does not filter the model unless converted by context transition.
Query folding is when Power Query translates transformation steps into a native source query, such as SQL, so the source does the work and only the result is loaded. It makes refreshes much faster and is required for incremental refresh and efficient DirectQuery. Steps like filtering, removing columns and grouping usually fold on SQL sources, while some custom M functions or adding an index column can break folding.
10
What is a disconnected table and how is it used for what-if analysis?
A disconnected table has no relationship to the model and is used to capture a user's choice through a slicer. For example, a Discount table from 0 to 30 percent created with GENERATESERIES, or the What-if parameter feature, lets a measure read SELECTEDVALUE(Discount[Value]) and calculate projected revenue. Since it does not filter facts, the measure decides how the selection is applied.
11
What is the grain of a fact table and why does it matter?
Grain is the level of detail of one fact row, such as one row per order line per day. It must be defined clearly before building measures, because mixing grains in one table, for example daily sales and monthly targets, causes double counting and wrong totals. Facts at different grains should be separate tables related to dimensions at the matching level.
12
How does the VertiPaq engine store and compress data?
VertiPaq is the in-memory columnar engine for Import models. It stores each column separately and compresses it using dictionary encoding, which maps unique values to integers, value encoding for numbers, and run-length encoding for repeated sequences. Compression depends mostly on column cardinality, so columns with few unique values compress very well, while unique IDs and precise timestamps take most of the memory.
Calculation groups are reusable sets of calculation items that modify any measure, such as Current, YTD, Previous Year and YoY %, using SELECTEDMEASURE(). Instead of writing YTD versions of 30 measures, you write one item that applies to all. They can be created in Power BI Desktop model view or Tabular Editor. They reduce measure sprawl but change how measures behave, so naming and format strings need care.
14
How do you handle a many-to-many relationship between sales and budget at different grains?
Budgets are often at month and category level while sales are daily and per product. Instead of a direct many-to-many link, relate both fact tables to shared dimensions at the right level: Date and a Category dimension. Budget relates to Category and to Date through a month-level key or the first day of month. Measures then compare them, and visuals should use only the shared grain.
15
What is a dataflow and how is it different from Power Query in Desktop?
A dataflow runs Power Query transformations in the Power BI Service and stores the result in the cloud, so multiple semantic models can reuse the same cleaned tables. Desktop Power Query logic lives inside one model only. Dataflows Gen2 in Fabric can also write output to lakehouses or warehouses. Dataflows centralise preparation logic, reduce duplicate source load and separate data prep from modelling.
16
What is Copilot in Power BI and what should you watch out for?
Copilot in Power BI uses generative AI to help create report pages, summarise visuals in narrative form, suggest DAX and answer questions about data in natural language. It needs supported capacity and tenant settings enabled by admins. Results depend heavily on a clean model with clear names, descriptions and synonyms, and outputs must be checked for correctness before sharing, especially generated DAX.
17
What is a decomposition tree and when is it useful?
The decomposition tree is an AI visual that lets users break down a measure across several dimensions in any order, for example splitting a sales drop by region, then category, then channel. It can also suggest the next split automatically using high value or low value analysis. It is useful for root-cause exploration by business users, though it is less suited to fixed reporting.
Incremental refresh partitions a large table by date so each refresh reloads only recent data, such as the last 10 days, while keeping older partitions, such as five years of history, unchanged. It uses the RangeStart and RangeEnd datetime parameters to filter the query, which must fold to the source. It cuts refresh time and load on sources, and can add a DirectQuery partition for real-time data.
Row-level security restricts which rows of data a user can see in a semantic model. You define roles in Desktop with DAX filter expressions on tables, such as Store[Region] = "North", then assign users or security groups to roles in the Service. Filters propagate through relationships to fact tables. RLS applies to viewers; workspace Admins, Members and Contributors can see all data.
20
How do you explain a technical limitation to a business user who wants something impossible?
I acknowledge the goal, explain the limitation in simple terms and offer an alternative that meets most of the need. For example, when a user wanted live updates every second on a DirectQuery report over a slow system, I explained the load it would create and offered a 15-minute refresh plus an alert for threshold breaches. They accepted because the key decision was covered.
21
What is a role-playing dimension and how do you handle it?
A role-playing dimension is one dimension used in several roles, such as Date acting as Order Date, Ship Date and Delivery Date. Options are one Date table with one active and other inactive relationships used through USERELATIONSHIP in measures, or separate copies of the date table for each role so users can slice by each date independently. The second is easier for users but increases model size slightly.
22
Refresh takes 90 minutes because of heavy Power Query steps. How do you speed it up?
I would check query folding with View Native Query or the folding indicators and move non-folding steps after folding ones. Filter rows and remove unused columns as early as possible, push joins and aggregations into SQL views, avoid row-by-row custom functions over large tables, and disable load for staging queries. For large fact tables, set up incremental refresh so only recent partitions reload.
23
Users want to analyse sales by both order date and delivery date in the same report. What do you do?
I would keep the Date table related to OrderDate as the active relationship and add an inactive relationship to DeliveryDate. Then create measures like Sales by Delivery Date using USERELATIONSHIP. If users need separate slicers for both dates at the same time, I would create a second Delivery Date table as a role-playing copy related actively to the delivery column.
24
How do you publish a report and share it with business users?
In Desktop, choose Publish and select a workspace. In the Service, set up the semantic model's credentials and scheduled refresh. Then create or update an app from the workspace, choose the reports and audiences, and grant access to user groups. Apps give consumers a clean, stable view while authors keep working in the workspace. Avoid sharing from My workspace for team content.
25
Users say the report shows yesterday's data even after a refresh. How do you troubleshoot?
I would check the refresh history in the Service for errors or whether the refresh actually ran, confirm the gateway is online and has valid credentials, and verify the source itself has the latest data. Then check for cached dashboard tiles, query caching, or an incremental refresh policy that excludes recent days. Also confirm users are viewing the right workspace or app version, which may need republishing.
26
Scheduled refresh fails every night with a gateway error. How do you troubleshoot?
I would check the refresh history error details, confirm the gateway machine is on and the gateway service is running and updated, and verify data source credentials in the gateway connection have not expired. Then test connectivity from the gateway machine to the source and check that the source names in the model exactly match the gateway data source. Gateway logs and network or firewall changes are next.
Deployment pipelines move content through stages, typically Development, Test and Production workspaces. Creators deploy reports and semantic models from one stage to the next, compare differences, and use deployment rules to change data source connections or parameters for each stage. This supports controlled releases and testing before users see changes. They work alongside Git integration for version control in Fabric workspaces.
Only one active relationship can exist between two tables along a filter path; additional relationships become inactive and appear as dashed lines. For example, Sales may have OrderDate and ShipDate both linked to Date, with OrderDate active. An inactive relationship is used only inside a measure with USERELATIONSHIP, such as CALCULATE([Sales], USERELATIONSHIP(Sales[ShipDate], 'Date'[Date])).
29
What is the difference between SELECTEDVALUE and VALUES?
SELECTEDVALUE(column, alternate) returns the single value of a column when exactly one is visible in the filter context, otherwise the alternate or blank. VALUES(column) returns a table of distinct visible values, including blank for unmatched rows, and can be used in iterators or as a filter. Use SELECTEDVALUE for titles and parameters, and VALUES when you need to loop over the visible items.
30
How do you use Performance Analyzer in Power BI Desktop?
Open View, Performance Analyzer, click Start recording and then Refresh visuals or interact with the page. Each visual shows total duration split into DAX query, visual display and other. Sort by duration to find the slowest visuals, expand them and choose Copy query to paste into DAX Studio for deeper analysis. Clear the cache between tests so timings are realistic.
31
Reports slow down every morning at 9 a.m. on shared capacity. What could be happening?
Morning peaks often combine many users opening reports with heavy scheduled refreshes running at the same time. I would check the Fabric Capacity Metrics app for throttling and identify the heaviest items. Fixes include moving refreshes to off-peak hours, optimising the largest semantic models, spreading workloads across capacities, or scaling the capacity. Communicating refresh windows to teams also helps.
32
Create a calculated column that categorises customers by their lifetime sales.
Customer Band =
VAR s = CALCULATE(SUM(Sales[Amount]))
RETURN
SWITCH(TRUE(),
s >= 500000, "Platinum",
s >= 100000, "Gold",
s > 0, "Silver",
"No Purchase")
This is a column on the Customer table; CALCULATE triggers context transition so each row gets its own customer's sales.
33
Users need to filter a report by 5,000 product names using a slicer, and it is slow. What do you do?
A slicer with thousands of values sends heavy queries and is hard to use. I would add a hierarchy so users pick category and subcategory first, enable the slicer search box, and consider a text filter visual for direct search. Apply buttons in query reduction settings stop queries on every click. If the slicer depends on a measure-filtered list, I would simplify that logic.
34
How do you configure incremental refresh on a sales table?
In Power Query create RangeStart and RangeEnd parameters of type Date/Time, and filter OrderDate so it is greater than or equal to RangeStart and less than RangeEnd. Confirm the step folds to the source. In Desktop right-click the table, choose Incremental refresh, set to archive, say, 5 years and refresh the last 10 days, optionally detect data changes. Publish, and the Service creates partitions on first refresh.
35
Users complain the report page is cluttered with 22 visuals. How do you redesign it?
I would ask which decisions the page supports and keep only visuals that answer those questions. Then group related KPIs into a top row of cards, keep two or three main charts, and move details to drill-through or tooltip pages. Slicers can move into a collapsible panel using bookmarks. Fewer visuals also make the page load faster.
36
Tell me about a time you handled a data access or security issue.
Situation: a sales user reported seeing another region's numbers after a reorganisation. Task: fix the leak fast. Action: I confirmed the security mapping table had stale entries, removed the user from the wrong row, refreshed the model and then automated the mapping from the HR system with a daily refresh. I informed the data owner. Result: access stayed correct after later reorganisations, and quarterly access reviews were added.
37
Explain how RANKX works and common mistakes with it.
RANKX iterates a table, evaluates an expression for each row and ranks the current value among them. A common pattern is RANKX(ALL('Product'[Product]), [Total Sales]). Mistakes include not using ALL, so every product ranks 1 because only one is visible, ranking over a table with duplicates, and forgetting ties handling with the Dense option. The newer RANK function supports multi-column ordering.
38
A colleague's measure uses FILTER(Sales, ...) on a 50-million-row table and is slow. How would you rewrite it?
I would replace FILTER over the whole fact table with column filters in CALCULATE, such as CALCULATE([Sales], Sales[Channel] = "Online"), which the engine can apply efficiently. If a condition uses a measure, iterate over a small dimension table instead, like FILTER(VALUES(Customer[CustomerID]), [Sales] > 10000). Then test in DAX Studio with server timings before and after.
39
What is cross-filter direction and why is bi-directional filtering risky?
Cross-filter direction controls which way filters travel across a relationship. Single direction, from the one side to the many side, is the default and safest. Bi-directional lets the fact table filter the dimension too, which can be useful for slicers showing only relevant values, but it can cause ambiguous paths, unexpected results and slower queries. Prefer single direction and use CROSSFILTER inside specific measures when needed.
40
Business wants year-over-year growth, but the current year is incomplete. How do you compare fairly?
Comparing a partial year with a full last year understates growth. I would limit last year to the same date range, for example using the last date with sales this year and filtering last year's dates up to the matching day, or compare year-to-date to previous year-to-date. I would label the visual clearly as YTD versus same period last year so readers understand it.
Frequently asked questions
What is asked in Power BI interviews for 3 years of experience?
Three-year candidates should own complete solutions. Questions move to incremental refresh, DAX Studio and VertiPaq Analyzer, calculation groups, deployment pipelines, composite models and handling large data. Expect scenario rounds on slow reports, refresh failures and conflicting numbers between departments, where your structured troubleshooting approach matters most. Some companies add a take-home case study on a sample dataset.
How many questions should I practise before a Power BI interview?
Work through all 40 questions here, then the company-wise sets for the employers you are targeting. Quality matters more than count: be able to explain why each answer is right.
Which topics matter most for 3 years of experience?
Focus first on Incremental refresh, DAX Studio tuning, Deployment pipelines, Troubleshooting scenarios.