A semantic model has been reviewed by the central data team and should be signposted to users as trustworthy and officially supported. What should be applied?

Certification through the endorsement settings

A sensitivity label of Highly Confidential

A refresh schedule of eight times per day

A workspace role of Viewer for the data team

A report must send an email to a group whenever a KPI on a dashboard tile crosses a threshold. Which capability provides this?

A data alert set on the dashboard tile

A calculation group in the model

A drillthrough page filtered to the KPI

A dataset must be shared so other authors can build their own reports on it without duplicating the model. What should be configured?

Build permission on the semantic model

A separate .pbix copy for each author

Reshare permission on each report

A new gateway cluster for the model

Row-level security has been defined with a role named RegionManager. What additional step is required before it takes effect for users?

Assign users or security groups to the role in the Power BI service

Convert the model to DirectQuery mode

Add the role name as a column in the fact table

Republish the report with the role deleted

Content in a workspace must carry a classification that follows exported files into Excel and PDF. Which capability provides that?

Sensitivity labels from Microsoft Purview Information Protection

Workspace endorsement set to Promoted

A data alert on each report page

A scheduled refresh notification

Finance users should receive a formatted snapshot of a report in their inbox every Monday morning without opening Power BI. What should be configured?

A report subscription with a weekly schedule

A field parameter for the report period

A semantic model refresh must run four times per day, but the workspace is on a shared capacity. What limitation should the analyst plan around?

Shared capacity caps scheduled refreshes per day, unlike dedicated capacity

Shared capacity disables row-level security

Shared capacity does not support scheduled refresh at all

Shared capacity requires DirectQuery for every model

A measure written as DIVIDE(SUM(Sales[Profit]), SUM(Sales[Revenue])) is preferred over the division operator. What is the practical advantage?

DIVIDE handles division by zero and returns a specified alternate result

DIVIDE forces the calculation to run in Power Query instead of DAX

DIVIDE automatically ignores all filter context

DIVIDE converts the result to a text data type

← All practice tests

PRACTICE ENGINE · PL-300

PL-300 Practice Test.
Free practice exam — 67 quality-checked questions, instant feedback.

Q: Are these questions like the real PL-300 items?

They are scenario-driven the same way: a requirement is described and you choose the feature that meets it. What they cannot reproduce are the interactive components Microsoft says the exam may include, where you work in an interface rather than pick an option. Microsoft's free exam sandbox exists for exactly that, so use both.

Q: How are these questions distributed across the exam objectives?

In proportion to Microsoft's published weights: roughly a quarter each on preparing data, modelling data, and visualizing and analyzing data, with the remainder on managing and securing Power BI. That last group is the one most candidates skip in practice and then lose marks on, because it is learned by administering a workspace rather than by building reports.

Q: What score on practice questions means I am ready?

There is no honest conversion, because the real cut score of 700 is scaled rather than a percentage and Microsoft does not publish how many questions the exam contains. A practical working target is consistently clearing 80 percent across all four objective areas, with no single area lagging badly, since the exam samples all four every time.

Q: Do I need Power BI installed to prepare properly?

Yes, and it is the highest-value hour of preparation you can spend. Microsoft states you should be proficient at Power Query and DAX, and neither is learned by reading. Build one small star schema with a marked date table, write a year-to-date and a prior-year measure, publish it, then add a row-level security role and assign yourself to it.

Q: How current are these questions with the exam objectives?

They follow the skills measured that took effect on 20 April 2026, which is the version currently published by Microsoft. Microsoft updates objectives periodically and publishes a change log with each revision, so check the official study guide before your exam date if some months have passed.

Written by Every Exam Prep Editorial TeamSource and review policy

Verified against the official content outline August 19, 2026

These are original study questions written from published exam objectives—not recalled, copied, or confidential live-exam items. Always confirm current coverage with the official sources linked on this page.

67 Questions100% FreeNo Signup

✓ No registration✓ No credit card✓ Start immediately

Difficulty

QUESTION 1 / 67Manage and secure Power BIEasy0/0

A curated set of reports and dashboards must be distributed to 300 people with a simple, consistent navigation experience. Which distribution method fits best?

0/0session

0 of 67 answered this set

Browse all questions & answers

1. A curated set of reports and dashboards must be distributed to 300 people with a simple, consistent navigation experience. Which distribution method fits best?
- A. Exporting every report to PDF weekly
- B. Adding all 300 people as workspace Members
- C. Publishing a Power BI app from the workspace
- D. Sending each person the .pbix file by email
Show answer & explanation
Answer: C
An app packages selected workspace content with its own navigation and audience-based access, which is designed for broad read-only distribution. Making everyone a Member grants editing rights, and file or PDF distribution loses interactivity and version control.
2. A semantic model has been reviewed by the central data team and should be signposted to users as trustworthy and officially supported. What should be applied?
- A. Certification through the endorsement settings
- B. A refresh schedule of eight times per day
- C. A workspace role of Viewer for the data team
- D. A sensitivity label of Highly Confidential
Show answer & explanation
Answer: A
Endorsement marks content as Promoted or Certified, and certification is restricted to reviewers designated by the tenant admin. Sensitivity labels classify confidentiality, and refresh frequency and roles say nothing about quality.
3. A report must send an email to a group whenever a KPI on a dashboard tile crosses a threshold. Which capability provides this?
- A. A calculation group in the model
- B. A bookmark applied to the tile
- C. A drillthrough page filtered to the KPI
- D. A data alert set on the dashboard tile
Show answer & explanation
Answer: D
Data alerts monitor card, gauge, and KPI tiles and notify when a value crosses a defined threshold. Bookmarks, calculation groups, and drillthrough operate inside the report and do not trigger notifications.
4. A dataset must be shared so other authors can build their own reports on it without duplicating the model. What should be configured?
- A. Build permission on the semantic model
- B. A separate .pbix copy for each author
- C. A new gateway cluster for the model
- D. Reshare permission on each report
Show answer & explanation
Answer: A
Build permission grants the right to create new content from an existing semantic model, which is the basis of the shared-model pattern. Copying the .pbix multiplies models and defeats the single source of truth.
5. Row-level security has been defined with a role named RegionManager. What additional step is required before it takes effect for users?
- A. Assign users or security groups to the role in the Power BI service
- B. Republish the report with the role deleted
- C. Add the role name as a column in the fact table
- D. Convert the model to DirectQuery mode
Show answer & explanation
Answer: A
Defining a role in Desktop only writes the filter expression; membership is managed in the service on the semantic model's security settings. Without members, the role filters nobody.
6. Content in a workspace must carry a classification that follows exported files into Excel and PDF. Which capability provides that?
- A. Sensitivity labels from Microsoft Purview Information Protection
- B. A scheduled refresh notification
- C. Workspace endorsement set to Promoted
- D. A data alert on each report page
Show answer & explanation
Answer: A
Sensitivity labels classify content and persist into supported export formats so protection travels with the file. Endorsement signals quality rather than confidentiality, and alerts and refresh notifications carry nothing into exports.
7. Finance users should receive a formatted snapshot of a report in their inbox every Monday morning without opening Power BI. What should be configured?
- A. A field parameter for the report period
- B. A data alert on a table visual
- C. An incremental refresh policy
- D. A report subscription with a weekly schedule
Show answer & explanation
Answer: D
Subscriptions email a rendered snapshot of a report page on a schedule the author defines. Alerts respond to thresholds rather than a calendar, and refresh policies and field parameters do not deliver anything by email.
8. A semantic model refresh must run four times per day, but the workspace is on a shared capacity. What limitation should the analyst plan around?
- A. Shared capacity caps scheduled refreshes per day, unlike dedicated capacity
- B. Shared capacity disables row-level security
- C. Shared capacity requires DirectQuery for every model
- D. Shared capacity does not support scheduled refresh at all
Show answer & explanation
Answer: A
Shared capacity allows scheduled refresh but limits how many times per day it may run, while dedicated capacity raises that ceiling substantially. Storage mode and security features are not affected by the capacity type.
9. An analyst must document which upstream tables feed a published report before making a breaking change. Which service feature shows those dependencies?
- A. The Q&A synonyms editor
- B. Lineage view for the workspace
- C. The Performance Analyzer pane
- D. The Selection pane in Desktop
Show answer & explanation
Answer: B
Lineage view maps data sources, semantic models, reports, and dashboards so downstream impact is visible before a change. Desktop panes describe a single file, and the synonyms editor tunes natural-language matching.
10. A measure written as DIVIDE(SUM(Sales[Profit]), SUM(Sales[Revenue])) is preferred over the division operator. What is the practical advantage?
- A. DIVIDE handles division by zero and returns a specified alternate result
- B. DIVIDE converts the result to a text data type
- C. DIVIDE automatically ignores all filter context
- D. DIVIDE forces the calculation to run in Power Query instead of DAX
Show answer & explanation
Answer: A
DIVIDE performs a safe division that returns blank or a supplied alternate value when the denominator is zero, avoiding infinity and error results. It does not change where the calculation runs or how filter context applies.
11. A measure must count how many distinct customers made a purchase in the selected period. Which DAX function is appropriate?
- A. COUNTBLANK(Sales[CustomerKey])
- B. DISTINCTCOUNT(Sales[CustomerKey])
- C. COUNTROWS(Sales)
- D. SUM(Sales[CustomerKey])
Show answer & explanation
Answer: B
DISTINCTCOUNT returns the number of unique values in a column within the current filter context, which is exactly a unique-customer count. COUNTROWS counts transactions, summing a key is meaningless, and COUNTBLANK counts missing values.
12. A report shows correct totals per product but the grand total does not equal the sum of the rows for a ratio measure. What is the usual explanation?
- A. The relationship between the tables is inactive
- B. The measure was written as a calculated column
- C. The measure is evaluated in the total row's filter context, not summed from rows
- D. The visual has a page-level filter applied
Show answer & explanation
Answer: C
A measure is recomputed for every cell including the total, so a ratio at the total level uses total numerator over total denominator rather than adding row ratios. This is expected behaviour, not a modelling error.
13. Which situation most clearly calls for a composite model that mixes Import and DirectQuery tables?
- A. The report must work with no data source connection at all
- B. All data already lives in a single Excel workbook
- C. Every table is small and changes only once per year
- D. Small dimensions can be imported while a huge, frequently updated fact table stays in DirectQuery
Show answer & explanation
Answer: D
Composite models let each table choose its own storage mode, so small slow-changing dimensions gain import speed while a massive live fact table stays queried at source. Uniformly small or static data has no need for the extra complexity.
14. A DAX measure must return sales for the current year to date. Which function combination is appropriate with a marked date table?
- A. COUNTROWS('Date') filtered to the current year
- B. SUM(Sales[Amount]) * 12
- C. AVERAGE(Sales[Amount]) over all rows
- D. CALCULATE(SUM(Sales[Amount]), DATESYTD('Date'[Date]))
Show answer & explanation
Answer: D
DATESYTD returns the set of dates from the start of the year through the last date in context, and CALCULATE evaluates the sum over that set. Multiplying or averaging does not respect period boundaries.
15. A report must be usable by people who navigate with a keyboard and a screen reader. Which combination of settings most directly supports that?
- A. Increasing the number of visuals per page
- B. Using color alone to distinguish every series
- C. Setting tab order and adding alt text to visuals
- D. Disabling the filter pane for all users
Show answer & explanation
Answer: C
Tab order controls the sequence a keyboard user moves through, and alt text gives a screen reader something meaningful to announce for each visual. Relying on color alone fails users with color vision deficiency, and denser pages make navigation harder.
16. An analyst wants Power BI to identify which categories most explain a sudden drop in revenue between two months. Which built-in capability does this?
- A. The Performance Analyzer pane
- B. The Query Diagnostics tool in Power Query
- C. The Analyze feature's Explain the decrease option
- D. The Manage Relationships dialog
Show answer & explanation
Answer: C
Right-clicking a data point and choosing Analyze runs an automated breakdown that ranks contributing categories for an increase or decrease. Performance Analyzer and Query Diagnostics measure speed, and Manage Relationships edits the model.
17. A scatter chart of customers by revenue and order count should highlight customers that behave very differently from the rest. Which analytic approach fits best?
- A. Anomaly or outlier detection combined with clustering
- B. Adding a running total to the revenue axis
- C. Sorting the customer names alphabetically
- D. Converting the visual to a single card
Show answer & explanation
Answer: A
Clustering groups points that behave similarly and makes the points that fall outside those groups stand out as outliers. Running totals, cards, and alphabetical sorting do not identify unusual behaviour in a two-measure space.
18. A pixel-perfect invoice that must print consistently across hundreds of pages is required alongside an interactive dashboard. Which Power BI artifact suits the invoice?
- A. A dashboard tile pinned from a visual
- B. A Q&A visual on the dashboard page
- C. A streaming dataset tile
- D. A paginated report
Show answer & explanation
Answer: D
Paginated reports are built for fixed layout and controlled pagination, which is what printed or exported multi-page documents require. Interactive report visuals, dashboard tiles, and streaming tiles are designed for on-screen exploration.
19. Two slicers for Year appear on separate pages and must always show the same selection. Which feature enforces that?
- A. Edit interactions
- B. The Selection pane
- C. Sync slicers
- D. Page-level filters
Show answer & explanation
Answer: C
Sync slicers link slicer selections across chosen pages so a change on one page carries to the others. Edit interactions controls how visuals affect each other on a single page, and page filters cannot be driven by a user's slicer choice elsewhere.
20. Selecting a bar in one visual is filtering three other visuals, but a stakeholder wants one of them left unfiltered. What should the author change?
- A. Delete the relationship between the underlying tables
- B. Add a row-level security role for that stakeholder
- C. Move the visual to a different report page
- D. Edit interactions and set that visual to None for the source visual
Show answer & explanation
Answer: D
Edit interactions sets, per visual pair, whether a selection filters, highlights, or does nothing. Breaking a model relationship would affect every calculation, and moving the visual or applying security solves a different problem.
21. A line chart of daily orders looks extremely noisy and the underlying trend is hard to see. Which analytics-pane addition best communicates the trend?
- A. A constant line set to zero
- B. A trend line or forecast on the analytics pane
- C. Data labels on every daily point
- D. A second y-axis with the same measure
Show answer & explanation
Answer: B
A trend line fits the series and makes direction visible through the noise, and forecasting extends that fit forward with a confidence band. Adding labels to every point or duplicating the axis increases clutter without revealing the trend.
22. Executives will consume a report primarily on phones. What should the author do to make the experience usable?
- A. Turn off all filters on the report
- B. Reduce the model to fewer than ten tables
- C. Export the report to a static image file
- D. Build a mobile layout with the phone view in Power BI Desktop
Show answer & explanation
Answer: D
Phone view lets the author arrange a separate portrait layout with the most important visuals sized for a small screen. Model size and filter availability are unrelated to how the report renders on a phone.
23. A user should be able to swap the measure shown in a chart between Revenue, Margin, and Units without the author building three visuals. Which approach fits?
- A. Three overlapping visuals toggled by page filters
- B. A row-level security role per measure
- C. A field parameter bound to a slicer
- D. A calculated column combining the three measures
Show answer & explanation
Answer: C
Field parameters expose a list of fields or measures that a slicer can switch, changing what a single visual displays. Stacking hidden visuals is fragile, and neither calculated columns nor security roles let users choose a measure.
24. A report page should let users type a plain-language question such as 'total sales by region last quarter' and get a visual back. Which feature provides this?
- A. The Key influencers visual
- B. The decomposition tree visual
- C. The Q&A visual
- D. The Performance Analyzer pane
Show answer & explanation
Answer: C
The Q&A visual parses natural-language questions against the semantic model and renders a matching visual. Key influencers and the decomposition tree are analytic visuals driven by field wells rather than typed questions.
25. A workspace admin must give a team the ability to publish and edit content in a workspace but not to add or remove other users. Which workspace role fits?
- A. Admin
- B. Contributor with no publish rights
- C. Viewer
- D. Member
Show answer & explanation
Answer: D
Members can publish, edit, and share content and can add others only at lower roles, while full user management belongs to Admins. Viewers can only consume content, so they cannot publish at all.
26. Sales managers must each see only the rows for their own region, using a single published report. Which feature enforces this?
- A. Row-level security roles with a DAX filter on the region table
- B. A bookmark saved for each manager
- C. Separate report pages hidden from other managers
- D. A page-level filter set to each manager's region
Show answer & explanation
Answer: A
Row-level security applies a DAX filter to the model itself and maps users to roles, so restricted rows never reach the client. Report filters, bookmarks, and hidden pages only change what is displayed and can be bypassed.
27. An imported semantic model reads from an on-premises SQL Server. What is required for scheduled refresh to succeed in the Power BI service?
- A. A paginated report subscription
- B. An on-premises data gateway configured for that data source
- C. Row-level security enabled on the model
- D. A Power BI Desktop instance left running on a workstation
Show answer & explanation
Answer: B
The service cannot reach a private network directly, so a gateway brokers the connection and credentials for refresh. Desktop is an authoring tool and is not part of scheduled refresh, and security and subscriptions are unrelated to connectivity.
28. A model consumes far more memory than expected. Which change most reliably reduces model size?
- A. Increasing the number of report pages
- B. Removing unused columns, especially high-cardinality ones
- C. Adding more calculated columns to precompute values
- D. Converting all measures into calculated columns
Show answer & explanation
Answer: B
The VertiPaq engine compresses column by column, and high-cardinality columns such as timestamps or GUIDs compress poorly, so removing unused ones frees the most memory. Adding calculated columns increases stored data rather than reducing it.
29. A report must show data from a 400-million-row fact table that is refreshed hourly in a SQL Server data warehouse, and users need current values without waiting for a dataset refresh. Which storage mode best fits this requirement?
- A. A live connection to a local Excel workbook
- B. Import mode with incremental refresh disabled
- C. Import mode with a scheduled daily refresh
- D. DirectQuery mode against the warehouse
Show answer & explanation
Answer: D
DirectQuery leaves the data in the source and issues queries at report time, so users always see current values and no copy of the 400 million rows is loaded into the model. Import mode would require moving and refreshing that volume, which does not meet the freshness requirement.
30. In Power Query, a column of order dates has been imported as text because the source system uses a regional format. Which action resolves this correctly for downstream time intelligence?
- A. Create a DAX calculated column that concatenates the text values
- B. Change the column data type to Whole Number
- C. Change the column data type to Date using a locale in the type change
- D. Leave the column as text and sort it alphabetically
Show answer & explanation
Answer: C
Using Change Type With Locale tells Power Query how to interpret the regional format so the conversion succeeds instead of producing errors. Time intelligence functions require a real Date or DateTime column, so text cannot be left in place.
31. A query in Power Query shows a column profile indicating that 12 percent of values in the CustomerID column are null. Which Power Query feature surfaced that percentage?
- A. Advanced Editor
- B. Query dependencies
- C. Data source settings
- D. Column quality
Show answer & explanation
Answer: D
Column quality reports the share of valid, error, and empty values for a column, which is where the null percentage appears. Query dependencies show relationships between queries, and the Advanced Editor exposes the M code rather than profiling statistics.
32. A sales table has one column per month, with twelve month columns holding revenue values. What Power Query transformation converts this into a tidy structure with one row per month?
- A. Transpose the entire table
- B. Unpivot the month columns
- C. Group by the month columns
- D. Pivot the month columns
Show answer & explanation
Answer: B
Unpivoting turns the selected column headers into values in an attribute column and their cell contents into a value column, producing one row per month per record. Pivoting does the opposite, and transposing swaps all rows and columns indiscriminately.
33. Two queries must be combined so the result contains all columns from a Sales query plus matching product attributes from a Products query. Which Power Query operation should be used?
- A. Transpose
- B. Merge queries
- C. Append queries
- D. Group by
Show answer & explanation
Answer: B
Merge joins two queries side by side on one or more key columns, widening the result with columns from the second query. Append stacks rows from queries that share a structure, which is not what adding attribute columns requires.
34. A developer wants the same file path used by six different queries to be changeable in one place before publishing. What is the appropriate Power Query construct?
- A. A parameter referenced by each query
- B. A bookmark stored in the report
- C. A separate workspace for each environment
- D. A calculated column added to each table
Show answer & explanation
Answer: A
Parameters hold a single value that queries reference, so changing the parameter updates every query that uses it. Calculated columns operate on loaded data and bookmarks capture report state, so neither can redirect a source path.
35. A query references another query to derive a filtered subset, and both are loaded to the model. What is the practical difference between using Reference and using Duplicate for this?
- A. Duplicate always produces a DirectQuery connection
- B. Duplicate and Reference behave identically in every respect
- C. Reference starts from the original query's output, so upstream edits flow through
- D. Reference copies all steps so later edits to the original are not inherited
Show answer & explanation
Answer: C
A referenced query takes the output of the source query as its starting point, so any change upstream propagates automatically. A duplicated query copies the step list, creating an independent chain that diverges once either side is edited.
36. A JSON column from a web API contains nested records for each order line. What Power Query action makes those nested fields usable as ordinary columns?
- A. Remove the column and rebuild it with a DAX measure
- B. Enable incremental refresh on the query
- C. Change the column data type to Text and split by comma
- D. Expand the record or list column into individual columns
Show answer & explanation
Answer: D
Expanding a record or list column promotes its nested fields into flat columns that the model can use. Treating structured JSON as text and splitting it is fragile, and DAX measures cannot reconstruct source structure that was never loaded.
37. A model must load only the last 24 months of a very large fact table, refreshing recent partitions while leaving older ones untouched. Which configuration achieves this?
- A. A report-level filter set to the last 24 months
- B. A row-level security role that filters by date
- C. A bookmark that hides older data on each page
- D. Incremental refresh with RangeStart and RangeEnd parameters
Show answer & explanation
Answer: D
Incremental refresh partitions the table by a date range defined with the reserved RangeStart and RangeEnd parameters and refreshes only recent partitions. Security roles, report filters, and bookmarks affect what is displayed, not what is stored or refreshed.
38. During refresh, several rows fail because a numeric column contains the text value 'N/A'. Which approach removes the failure without discarding valid rows?
- A. Change the column type to Text and leave it unconverted
- B. Remove the entire column from the query
- C. Replace the offending values before changing the column data type
- D. Disable the query's load to the model
Show answer & explanation
Answer: C
Replacing the non-numeric placeholder with null or zero before the type change lets the conversion succeed for every row. Dropping the column loses data, and leaving it as text blocks aggregation downstream.
39. A star-schema model needs a table that supplies continuous dates for time intelligence, including dates with no sales. What is the recommended approach?
- A. Sort the fact table ascending by transaction date
- B. Use the date column on the fact table directly
- C. Create a dedicated date table marked as the model's date table
- D. Create a measure that generates dates at query time
Show answer & explanation
Answer: C
Time intelligence functions require an unbroken sequence of dates, which a fact table cannot guarantee because it only contains dates on which transactions occurred. A dedicated date table marked as such gives DAX the contiguous calendar it needs.
40. A relationship between a Product dimension and a Sales fact table is created. What cardinality and cross-filter direction are normally correct for a star schema?
- A. One-to-one with single direction
- B. One-to-many from Product to Sales with single direction
- C. Many-to-one from Product to Sales with both directions
- D. Many-to-many with both directions
Show answer & explanation
Answer: B
A dimension holds one row per product and the fact table holds many rows per product, which is one-to-many. Single-direction filtering from the dimension to the fact table is the default because it keeps filter propagation predictable and avoids ambiguity.
41. A Sales table has both an OrderDate and a ShipDate, and both must be usable for time intelligence against the same date table. Which modelling technique handles this?
- A. Create an inactive relationship and activate it with USERELATIONSHIP in a measure
- B. Create two active relationships to the same date table
- C. Duplicate the entire Sales table for the second date
- D. Merge OrderDate and ShipDate into one column
Show answer & explanation
Answer: A
Only one relationship between two tables can be active at a time, so the second date relationship is created as inactive. A measure wrapped in CALCULATE with USERELATIONSHIP activates it for that calculation only, which is the role-playing dimension pattern.
42. Which DAX expression correctly returns total sales for the same period one year earlier, given a proper date table?
- A. SUMX(Sales, Sales[Amount] * 0.5)
- B. SUM(Sales[Amount]) - 365
- C. CALCULATE(SUM(Sales[Amount]), SAMEPERIODLASTYEAR('Date'[Date]))
- D. FILTER(Sales, Sales[Date] < TODAY())
Show answer & explanation
Answer: C
SAMEPERIODLASTYEAR shifts the current filter context on the date column back by one year, and CALCULATE applies that shifted context to the aggregation. Subtracting a number from a sum or filtering rows arbitrarily does not shift the time context.
43. What is the essential difference between a calculated column and a measure in a Power BI semantic model?
- A. A calculated column is stored per row at refresh; a measure is evaluated at query time in filter context
- B. A measure is stored per row at refresh; a calculated column is evaluated at query time
- C. Calculated columns can only be written in Power Query M
- D. Both are stored in the model and consume identical memory
Show answer & explanation
Answer: A
A calculated column materializes a value for every row and occupies memory in the model, so it is computed during refresh. A measure carries no stored values and is recomputed for whatever filter context the visual supplies, which is why measures usually scale better.
44. A measure must return total sales ignoring any filter applied on the Product table. Which DAX function accomplishes this inside CALCULATE?
- A. VALUES(Product[Category])
- B. REMOVEFILTERS(Product)
- C. KEEPFILTERS(Product)
- D. RELATED(Product[Category])
Show answer & explanation
Answer: B
REMOVEFILTERS clears filters from the specified table so the calculation is evaluated as though no product filter existed. KEEPFILTERS does the opposite by preserving existing filters, and RELATED and VALUES retrieve values rather than modify context.
45. A model contains an inventory snapshot that should not be summed across dates but should be summed across warehouses. What kind of measure is required?
- A. A fully additive measure using plain SUM
- B. A non-additive measure that returns a constant
- C. A calculated column that concatenates warehouse and date
- D. A semi-additive measure such as one using LASTDATE or CLOSINGBALANCEMONTH
Show answer & explanation
Answer: D
Balances and stock levels are semi-additive: adding them across time double-counts, while adding them across other dimensions is valid. DAX handles this with functions that pick a point in time, such as LASTDATE or the closing-balance family.
46. Performance Analyzer shows one table visual taking 8 seconds while others render in under 200 milliseconds. What is the most appropriate first diagnostic step?
- A. Delete the visual and rebuild the entire report page
- B. Copy the visual's DAX query and analyze it in DAX query view
- C. Switch the whole model from Import to DirectQuery
- D. Increase the refresh frequency of the semantic model
Show answer & explanation
Answer: B
Performance Analyzer exposes the exact query behind each visual, and running it in DAX query view isolates whether the cost is in the measure logic or the rendering. Rebuilding the report or changing storage mode are large changes made before the cause is known.
47. A calculation group is added to a model containing calculation items named Current, Year to Date, and Prior Year. What does this achieve?
- A. It applies a reusable set of time calculations to many measures without duplicating them
- B. It permanently replaces the existing measures in the model
- C. It creates one physical table per calculation item at refresh
- D. It converts the model from Import to DirectQuery mode
Show answer & explanation
Answer: A
Calculation groups let a single set of items modify any measure a user selects, which avoids writing year-to-date and prior-year variants of every measure. They add expression logic, not stored tables, and they do not alter storage mode.
48. A Sales table stores an integer ProductKey and the Product table stores the same key. Why is joining on this integer key preferable to joining on the product name?
- A. Product names are always unique across every dimension
- B. Integer keys automatically enforce many-to-many cardinality
- C. Integer keys compress better and produce faster relationship scans
- D. Text columns cannot be used in relationships at all
Show answer & explanation
Answer: C
Narrow integer keys compress efficiently in VertiPaq and are cheaper to scan than wide text values, so relationships evaluate faster. Names are also prone to duplication and formatting differences, which makes them unreliable keys.
49. A report author needs to show the contribution of each region to total sales as parts of a whole, with six regions. Which visual is most appropriate?
- A. A stacked bar chart or treemap showing share of total
- B. A scatter chart with region on both axes
- C. A card visual showing the grand total only
- D. A line chart with region on the axis
Show answer & explanation
Answer: A
Part-to-whole comparisons need a visual that encodes share, which stacked bars and treemaps do while remaining readable with several categories. Line charts imply a sequence, scatter charts compare two measures, and a card cannot show composition.
50. Users must be able to click a product in a summary table and jump to a detail page already filtered to that product. Which report feature provides this?
- A. Drillthrough configured on the detail page
- B. A slicer synced across both pages
- C. A bookmark that hides the summary visuals
- D. Conditional formatting on the product column
Show answer & explanation
Answer: A
Drillthrough is configured on the destination page with the fields that scope it, and right-click navigation carries the selected context there. Synced slicers keep filters aligned but do not perform the navigation, and bookmarks change state rather than pass context.
51. A report page must show cells in a table shaded from light to dark according to profit margin. Which feature applies this?
- A. The Selection pane grouping of visuals
- B. A sync slicer on the margin column
- C. Conditional formatting with a background color scale
- D. A custom report theme applied to the whole file
Show answer & explanation
Answer: C
Conditional formatting maps a field's values to a color gradient applied to cell backgrounds, font, or data bars. Themes set global palettes, the Selection pane manages layering, and slicers filter rather than format.
52. A stakeholder wants to save a specific combination of filters, visual visibility, and sort order and return to it with one click. Which feature should be used?
- A. A row-level security role
- B. A drillthrough page
- C. An aggregation table
- D. A bookmark, optionally attached to a button
Show answer & explanation
Answer: D
Bookmarks capture report state including filters, selection, visibility, and sort, and a button can be wired to apply one. Drillthrough navigates with context, aggregations tune performance, and security roles restrict data.
53. A model uses a single flattened table containing product, customer, and date attributes alongside every transaction. Which drawback is most likely?
- A. Redundant repeated attribute values inflate model size and slow filtering
- B. DAX measures cannot be written against a single table
- C. Time intelligence functions are always faster on flat tables
- D. Relationships become mandatory even with one table
Show answer & explanation
Answer: A
Flattening repeats every descriptive attribute on every transaction row, which compresses worse than a star schema and makes filter evaluation more expensive. A star schema keeps attributes in small dimensions that filter a narrow fact table.
54. Which Power Query step order is most efficient when a source table has 200 columns but only 8 are needed and only 2024 rows are relevant?
- A. Remove unneeded columns and filter rows as early as possible
- B. Sort the table first, then remove columns last
- C. Add all calculated columns before any filtering
- D. Load everything, then remove columns after loading to the model
Show answer & explanation
Answer: A
Filtering and column removal early let query folding push the work to the source, so far less data crosses the wire and later steps operate on a smaller table. Deferring these steps forfeits folding and moves unnecessary data.
55. An analyst notices that the View Native Query option is greyed out after a certain Power Query step. What does this indicate?
- A. Query folding has stopped at that step, so later work happens in the mashup engine
- B. The step contains a syntax error in M
- C. The data source has become unavailable
- D. The query has been converted to DirectQuery
Show answer & explanation
Answer: A
View Native Query is available only while Power Query can translate the accumulated steps into a source query. When it greys out, folding has broken and everything downstream is processed locally, which usually costs performance.
56. A dataflow is created so several semantic models can reuse the same cleaned customer table. What is the main benefit of this approach?
- A. Row-level security is applied to all models automatically
- B. The transformation logic is authored once and reused across models
- C. Every model automatically switches to DirectQuery
- D. The gateway becomes unnecessary for on-premises sources
Show answer & explanation
Answer: B
A dataflow centralizes the Power Query logic in the service so multiple models consume one prepared table instead of each repeating the cleanup. It does not change storage mode, apply security, or remove the need for a gateway.
57. A dataset owner wants privacy levels configured so that data from a public web source is never combined with a confidential internal database. Where is this configured?
- A. In the visual's format pane
- B. In the model's calculation groups
- C. In the data source settings, by assigning privacy levels per source
- D. In the report's page-level filters
Show answer & explanation
Answer: C
Privacy levels are set per data source and govern whether the engine may fold data from one source into a query against another. Report filters, formatting, and calculation groups have no role in that isolation.
58. A shared semantic model already exists in the service and a new report should use it without importing data again. Which connection type is used?
- A. A live connection to the published semantic model
- B. An Excel workbook connection
- C. A composite model with all tables set to Import
- D. An import connection to the same source database
Show answer & explanation
Answer: A
A live connection points a new report at an existing published model, so there is one copy of the data and one set of measures. Re-importing recreates the model and splits the single source of truth.
59. A visual calculation is added directly to a matrix to show each row's percentage of the column total. What distinguishes a visual calculation from a model measure?
- A. It is stored as a column in the underlying fact table
- B. It runs during dataset refresh rather than at query time
- C. It can only be used in card visuals
- D. It is defined on the visual and operates over the visual's result set
Show answer & explanation
Answer: D
Visual calculations are written against the data already displayed in a visual, which makes running totals and share-of-total expressions much simpler than equivalent model DAX. They are not stored in the model and do not run at refresh.
60. An analyst wants to explain which factors are associated with high customer churn in a dataset. Which visual is designed for that question?
- A. A slicer with a dropdown
- B. A ribbon chart
- C. The key influencers visual
- D. A gauge visual
Show answer & explanation
Answer: C
The key influencers visual runs an automated analysis that ranks factors associated with a chosen outcome and quantifies their effect. Gauges show one value against a target, ribbon charts show rank change over time, and slicers filter.
61. A decomposition tree is used on a revenue measure with dimensions for region, channel, and product. What does the visual's high-value path do?
- A. It restricts the visual to a single dimension
- B. It removes all filters applied by other visuals
- C. It converts the measure into a calculated column
- D. It automatically expands the branch with the largest contribution at each level
Show answer & explanation
Answer: D
The high-value path option lets the visual choose the next level's largest contributor automatically, walking the drill path toward the biggest driver. It does not alter filters, the model, or the available dimensions.
62. A report theme JSON file is applied to standardize colors and fonts. Which statement about themes is accurate?
- A. A theme can only be applied in the Power BI service
- B. A theme sets default formatting that individual visuals can still override
- C. A theme changes the underlying semantic model structure
- D. A theme permanently locks every visual's formatting
Show answer & explanation
Answer: B
Themes supply default colors, fonts, and visual styling across a report while leaving per-visual overrides available. They are applied in Desktop and affect presentation only, never the model.
63. Users report that a matrix showing 40,000 rows is slow and hard to read. Which redesign best addresses both problems?
- A. Increase the visual's width to show more rows at once
- B. Duplicate the matrix on a second page
- C. Aggregate to a summary level and provide drillthrough to detail
- D. Add data labels to every row of the matrix
Show answer & explanation
Answer: C
Summarizing reduces the number of rows the engine must return and gives readers a scannable overview, while drillthrough keeps the detail one click away. Widening, labeling, or duplicating the visual does not reduce the returned row count.
64. An organization requires that a report page automatically show the newest data every 15 minutes on a wall display without user interaction. Which setting supports this for a DirectQuery model?
- A. A bookmark cycled by a button press
- B. Automatic page refresh configured on the report page
- C. A field parameter bound to a slicer
- D. A sensitivity label applied to the report
Show answer & explanation
Answer: B
Automatic page refresh re-queries visuals on a page at a defined interval and is supported for DirectQuery sources. Bookmarks, parameters, and labels do not re-run queries on a timer.
65. A published report must remain available to viewers while the author tests changes to the same content. Which practice supports this safely?
- A. Edit the published report directly in the production workspace
- B. Delete the report and republish it after each change
- C. Use deployment pipelines or separate development and production workspaces
- D. Ask viewers to stop opening the report during testing
Show answer & explanation
Answer: C
Separating development from production, whether through pipelines or distinct workspaces, lets changes be validated before they reach consumers. Editing or deleting production content exposes viewers to broken intermediate states.
66. A dashboard must combine tiles from three different semantic models into one view. Which Power BI artifact supports this?
- A. A dashboard, because tiles can be pinned from multiple reports and models
- B. A dataflow, which stores tiles rather than tables
- C. A single report page, which can connect live to only one model
- D. A paginated report, which supports one dataset only
Show answer & explanation
Answer: A
Dashboards are collections of pinned tiles and can draw from many reports and models at once. A report with a live connection binds to a single model, and dataflows store prepared tables, not visuals.
67. An analyst must ensure that a semantic model's scheduled refresh does not fail because a credential has expired. What is the appropriate maintenance action?
- A. Add more visuals to the report page
- B. Update the stored data source credentials on the semantic model settings
- C. Re-create the report from scratch in Desktop
- D. Switch the workspace to a Viewer-only membership
Show answer & explanation
Answer: B
Refresh uses credentials stored against the semantic model or gateway data source, so an expired secret is fixed by updating them in the service. Rebuilding reports or changing membership does not restore connectivity.

Explore more Microsoft Power Platform certifications

Family overview →

More in this category

2026 statistics

Key facts: PL-300 exam

700 of 1000

To pass

1h 40m

Time limit

$165

Exam fee

The PL-300 is administered by Microsoft, with a 1 hour 40 minutes time limit and a 700 of 1000 result.

This free PL-300 practice test has 67 original questions written to Microsoft's official content outline, last checked against it on August 19, 2026. Every question shows a worked explanation, and nothing here requires a signup.

As of 2026, the PL-300 exam fee is $165 (US list price; Microsoft prices the exam by the country or region it is proctored in).

Study by section weight

The cheat sheet is built like the exam blueprint

Open cheat sheet →

Every free resource for this exam

PL-300 exam guide PL-300 study guide PL-300 cheat sheet

Get a free PL-300 study plan

A week-by-week plan plus new practice questions, straight to your inbox.

Official sources

Primary documents used to verify the exam details shown on this page.

Study guide for Exam PL-300: Microsoft Power BI Data AnalystMicrosoft Learnlearn.microsoft.com
Microsoft Certified: Power BI Data Analyst Associate — Exam PL-300Microsoft Learnlearn.microsoft.com
Microsoft Certifications — Exam frequently asked questionsMicrosoft Learnlearn.microsoft.com

Last verified against the official exam content outline: August 19, 2026

Frequently asked questions

Are these questions like the real PL-300 items?