Every Exam PrepFREE EXAM PREP
Ask AI
STUDY GUIDE · PL-300

Microsoft Certified: Power BI Data Analyst Associate (Exam PL-300) Study Guide

Verified against the official content outline 6 sections
Written by Every Exam Prep Editorial TeamSource and review policyPublished August 19, 2026
Time limit
1h 40m
Passing score
700 of 1000
Exam fee
$165
Governing body
Microsoft

100 minutes. A scaled passing score of 700. A US$165 typical fee for an associate exam, priced by the country or region where you are proctored, with tax possibly on top. A 12-month certification life, renewed free by an online assessment on Microsoft Learn rather than by repaying for the exam. And a 24-hour wait before your first retake if it goes badly.

One number is missing on purpose: Microsoft does not publish how many questions are on PL-300. Every article that tells you it is 40 to 60 is guessing. That matters practically — you cannot budget minutes per question against a number nobody has confirmed — so pace yourself by the clock and by how much of the exam remains on the progress indicator instead.

The 700 is also not a percentage. It is a scaled score across a fixed range, so you cannot convert it to a number of questions you are allowed to miss.

Four skill areas, weighted as of the 20 April 2026 objectives update: Prepare the data at 25-30 percent, Model the data at 25-30 percent, Visualize and analyze the data at 25-30 percent, and Manage and secure Power BI at 15-20 percent.

Read those weights carefully before you plan. The three build areas are equal, which means modelling is worth as much as report building — and modelling is where most self-taught Power BI users are weakest, because you can produce good-looking reports for years without ever needing a proper star schema, a marked date table, or USERELATIONSHIP. If your day job is dashboards, this is where your gap is.

The fourth area is the one people skip and then fail on. Workspace roles, apps, build permission, gateways, endorsement, sensitivity labels, row-level security, and refresh limits are up to a fifth of the exam. None of it is hard, but almost none of it is learned by accident if you have never administered a tenant.

Microsoft also states that most questions cover generally available features, and that Preview features may appear when they are commonly used. Copilot-driven report authoring now sits in the published objectives, so treat it as in scope rather than as a novelty.

Microsoft states plainly that you should be proficient at using Power Query and Data Analysis Expressions. Those two tools cut across every skill area, and almost every scenario question resolves to one of them.

For Power Query, the concept that earns the most marks is query folding — the engine translating your steps into a native source query. Know that removing columns and filtering rows early preserves folding while some transformations break it, and know that a greyed-out View Native Query option is how you spot the break. Around that, be fluent in merge versus append, pivot versus unpivot, reference versus duplicate, parameters, privacy levels, and the RangeStart and RangeEnd parameters that incremental refresh requires by name.

For DAX, the whole subject is filter context. A measure is recomputed for every cell it appears in, which is why a ratio's grand total is not the sum of the rows above it, and CALCULATE is the function that modifies that context. Build outward from there: time intelligence over a marked date table, USERELATIONSHIP for role-playing dimensions, REMOVEFILTERS and KEEPFILTERS, DIVIDE for safe division, and the difference between a measure and a calculated column in both memory and evaluation time.

Work from the official study guide outward. It is free, it lists every bullet Microsoft assesses, and it carries a change log showing what moved at the last update. Treat each bullet as a checkbox you can either demonstrate in Power BI Desktop or cannot — that self-audit is more useful than any question dump.

Build one real model rather than reading about many. Import a few tables, shape them in Power Query, wire a proper star schema with a marked date table, and then write the measures: a base sum, a year-to-date, a prior-year comparison with a role-playing date relationship, a safe ratio, a distinct customer count. Anything you can build once you can recognize in a scenario question.

Then close the administration gap deliberately, because it will not close itself. Publish that model to a workspace. Create a role with a DAX filter and assign a member. Publish an app and look at what a Viewer sees. Configure a refresh and see what it demands. Look at Lineage view. An hour of this is worth more than a chapter of reading about it.

Use the free practice assessment and the exam sandbox before the day. The sandbox exists so the interface and the interactive question types are not new to you when the clock is running.

The exam is proctored and may include interactive components, so expect at least some items that ask you to work in an interface rather than pick a letter. Scheduling runs through Pearson VUE, and you can sit it at a test centre or online.

Register with a personal Microsoft account rather than a work or school account. Microsoft warns that records tied to an organizational account are lost and unrecoverable if you leave that organization, and this is the single most avoidable long-term mistake candidates make.

The exam is offered in ten languages: English, Japanese, Chinese (Simplified), Korean, German, French, Spanish, Portuguese (Brazil), Chinese (Traditional), and Italian. If it is not available in your preferred language you can request an additional 30 minutes. Note also that localized versions are updated roughly eight weeks after the English version, so soon after an objectives change the two can differ slightly.

With 100 minutes and no published item count, keep moving. Flag anything that needs a second look rather than stalling, and leave nothing unanswered — there is no published penalty for a wrong answer, so a considered guess strictly beats a blank.

You need 700 or greater to pass, and the score report breaks performance down by skill area, which tells you where to concentrate if you have to sit it again. A first retake is available 24 hours after the first attempt; the wait for subsequent retakes is longer.

If you passed, put a reminder in your calendar now. The credential is valid for 12 months and Microsoft associate certifications expire unless renewed, but renewal is a free online assessment on Microsoft Learn rather than a repeat of the paid exam. The renewal window opens well before expiry, and letting it lapse means paying US$165 again for something that would have cost nothing.

Keep an eye on the study guide's change log too. Microsoft updates objectives periodically — the current set took effect on 20 April 2026 — and the renewal assessment tracks the current version, not the one you originally sat.

PL-300 flashcards

33 cards on the highest-yield terms and rules. Grading uses spaced repetition and saves in this browser.

Card 1 of 330 mastered
Say the answer out loud before flipping.
Browse all 33 cards
  1. How long do you get for the PL-300 exam?

    Microsoft states you will have 100 minutes to complete the assessment. It is proctored and may include interactive components.

  2. What score do you need to pass PL-300?

    700 or greater on Microsoft's scaled range. That is a scaled cut score, not a percentage of questions answered correctly.

  3. How many questions are on the PL-300 exam?

    Microsoft does not publish a count for PL-300. Any specific figure you see quoted is third-party estimate, not an official number.

  4. What are the four skill areas and their weights?

    Prepare the data 25-30%, Model the data 25-30%, Visualize and analyze the data 25-30%, and Manage and secure Power BI 15-20%, as of the 20 April 2026 skills update.

  5. How much does PL-300 cost?

    Associate exams such as PL-300 typically cost US$165, priced by the country or region where the exam is proctored, and local taxes may apply.

  6. How often must the certification be renewed?

    Every 12 months. Microsoft associate certifications expire annually and renew free by passing an online assessment on Microsoft Learn.

  7. How soon can you retake PL-300 after failing?

    24 hours after the first attempt. Waiting periods for later retakes are longer and vary.

  8. When should you choose DirectQuery over Import?

    When the source is too large to load or must be current at report time. Import gives faster queries but only shows data as of the last refresh.

  9. What is query folding and how do you check it?

    Power Query translating your steps into a native source query. Right-click a step and check View Native Query — when it greys out, folding has stopped and the rest runs locally.

  10. What is a star schema and why does it matter?

    A narrow fact table joined one-to-many to descriptive dimension tables. It compresses better and filters faster than a flattened table that repeats attributes on every transaction row.

  11. Calculated column or measure?

    A calculated column is computed row by row at refresh and stored in the model; a measure is evaluated at query time in filter context and stores nothing. Prefer measures unless you need the value for slicing or relationships.

  12. What does CALCULATE actually do?

    It evaluates an expression under a modified filter context. Almost every non-trivial measure — year to date, prior year, share of total, ignore a slicer — is built this way.

  13. How do you handle a role-playing date dimension?

    Create the second relationship as inactive and activate it inside a measure with USERELATIONSHIP in CALCULATE. Only one relationship between two tables can be active at a time.

  14. Why does a ratio measure's grand total not equal the sum of the rows?

    Because a measure is recomputed in each cell's filter context, including the total row. The total is total numerator over total denominator, not the sum of row ratios.

  15. What is a semi-additive measure?

    One that sums correctly across some dimensions but not across time, such as an inventory or account balance. Use point-in-time DAX like LASTDATE or the closing-balance functions.

  16. Why use DIVIDE instead of the slash operator?

    DIVIDE performs a safe division that returns blank or a supplied alternate result when the denominator is zero, avoiding infinity and error values in visuals.

  17. What does a marked date table give you?

    A contiguous calendar with every date present, which time intelligence functions require. A fact table only contains dates on which transactions happened, so gaps break the calculations.

  18. What are RangeStart and RangeEnd for?

    They are the reserved date-time parameters that define an incremental refresh policy's partition range, so only recent partitions are refreshed on a very large table.

  19. How do you reduce a model that uses too much memory?

    Remove unused columns, especially high-cardinality ones like timestamps and GUIDs that compress poorly. Adding calculated columns makes the problem worse, not better.

  20. What is a calculation group for?

    Applying reusable logic — year to date, prior year, percentage change — to whatever measure the user selects, instead of authoring a time variant of every measure in the model.

  21. Drillthrough or bookmark?

    Drillthrough navigates to a detail page carrying the selected row's context. A bookmark restores a saved report state — filters, selection, visibility, sort — usually behind a button.

  22. What does Edit interactions control?

    Whether a selection in one visual filters, highlights, or does nothing to each other visual on the same page. It is set per pair of visuals.

  23. How do you let users switch which measure a chart shows?

    Create a field parameter listing the measures and put it in a slicer. The visual then displays whichever field the user selects, without duplicating visuals.

  24. Which report settings matter most for accessibility?

    Setting tab order for keyboard navigation and adding meaningful alt text to visuals. Never rely on color alone to distinguish series.

  25. Row-level security versus a report filter?

    RLS applies a DAX filter in the model and maps users to roles, so restricted rows never reach the client. A report filter only changes what is shown and can be worked around.

  26. What extra step does RLS need after publishing?

    Assign users or security groups to the role in the Power BI service. A role defined in Desktop filters nobody until it has members.

  27. When is an on-premises data gateway required?

    Whenever the service must reach a data source inside a private network — for scheduled refresh of an imported model or for DirectQuery against on-premises data.

  28. How should a report reach 300 read-only consumers?

    Publish a Power BI app from the workspace. Adding everyone as a workspace Member would grant editing rights instead of read-only access.

  29. Endorsement or sensitivity label?

    Endorsement (Promoted or Certified) signals quality and trust; Certified is restricted to designated reviewers. A sensitivity label signals confidentiality and persists into exported files.

  30. What does Build permission allow?

    Creating new reports and other content from an existing shared semantic model — the permission that makes one governed model serve many downstream reports.

  31. Data alert or subscription?

    A data alert fires when a dashboard card, gauge, or KPI tile crosses a threshold. A subscription emails a rendered report snapshot on a schedule you define.

  32. How do you find what a breaking model change will affect?

    Open Lineage view for the workspace. It maps sources, dataflows, semantic models, reports, and dashboards and the dependencies between them.

  33. How do you diagnose one slow visual?

    Run Performance Analyzer, copy that visual's DAX query, and run it in DAX query view to see whether the cost is in the query or in rendering.

PL-300 glossary

30 terms the PL-300 tests, defined in plain English.

Bookmark
A saved capture of report state including filters, selections, sorting, and visual visibility, which a button can reapply in one click. Bookmarks change presentation only and never restrict which data a user can reach.
Build permission
The permission that allows someone to create new reports and other content from an existing semantic model. It is what makes the shared-model pattern work: one governed model, many downstream reports, no duplicated copies.
CALCULATE
The DAX function that evaluates an expression under a modified filter context. Nearly every non-trivial measure — year-to-date, prior year, share of total, ignoring a slicer — is built by wrapping an aggregation in CALCULATE with filter arguments.
Calculated column
A DAX expression evaluated row by row during refresh and stored in the model like any other column. It consumes memory, so a measure is usually preferable unless the value is genuinely needed for slicing, grouping, or a relationship.
Calculation group
A model object containing calculation items that apply reusable logic — year to date, prior year, percentage change — to whatever measure a user selects. It removes the need to author a separate time variant of every measure in the model.
Composite model
A model that mixes storage modes table by table, typically importing small slow-changing dimensions while leaving a huge, frequently updated fact table in DirectQuery. It trades some complexity for a balance of freshness and speed.
Dataflow
Power Query logic authored and executed in the Power BI service, producing prepared tables that multiple semantic models can reuse. It moves shared cleanup out of individual files so the transformation is written once.
DAX
Data Analysis Expressions, the formula language used for measures, calculated columns, calculated tables, and security filters in a Power BI model. Its central idea is filter context: the same measure returns different results in different visual cells.
DirectQuery
A storage mode that leaves data at the source and issues queries at report time, so users always see current values and very large tables never have to be loaded. Query speed then depends on the source system and network.
Drillthrough
A navigation feature configured on a destination report page that receives the context of the row a user right-clicked. It lets a compact summary page stay readable while full detail sits one click away.
Endorsement
The Promoted and Certified labels that signal how trustworthy an item is. Certification is restricted to reviewers designated by the tenant administrator, which makes it a governance signal rather than something any author can apply.
Field parameter
A model object exposing a list of fields or measures that a slicer can switch, changing what a single visual displays. It replaces the older workaround of stacking several hidden visuals and toggling them with bookmarks.
Filter context
The set of filters in effect when a DAX expression is evaluated, contributed by slicers, visual axes, the filter pane, and security roles. CALCULATE exists to modify this context, which is what makes time intelligence and share-of-total measures possible.
Import mode
A storage mode that loads and compresses a copy of the source data into the model, giving the fastest query performance. Data is only as current as the last refresh, so it suits sources that change on a predictable schedule.
Incremental refresh
A refresh policy that partitions a large table by a date range defined with the reserved RangeStart and RangeEnd parameters, then refreshes only recent partitions. It keeps refresh windows short on tables too large to reload in full.
Lineage view
A workspace view that maps data sources, dataflows, semantic models, reports, and dashboards and the dependencies between them. It is the fastest way to see what downstream content a breaking model change would affect.
Measure
A DAX calculation evaluated at query time against whatever filter context a visual supplies, storing no values in the model. This is why a ratio measure's grand total is recomputed at the total level rather than summed from the rows above it.
On-premises data gateway
The service that lets the Power BI cloud reach data sources inside a private network for scheduled refresh and DirectQuery. Without it, an imported model built on a local SQL Server cannot refresh once published.
Paginated report
A fixed-layout report type built for printing and multi-page export, where pagination and page breaks are controlled precisely. It suits invoices and statements, while interactive Power BI reports suit on-screen exploration.
Performance Analyzer
A Power BI Desktop pane that records how long each visual takes to render and exposes the DAX query behind it. Copying that query into DAX query view isolates whether a slow visual is limited by the query or by rendering.
Power BI app
A packaged, read-only distribution of selected workspace content with its own navigation and audience-based access. It is the standard way to publish a curated set of reports to a large audience without granting workspace edit rights.
Power Query
The data preparation engine behind Get Data, where connections, transformations, and type changes are authored as an ordered list of steps in the M language. All shaping happens here before data reaches the model.
Query folding
Power Query's ability to translate transformation steps into a single native query executed by the source system. Folding keeps work close to the data, and it stops at the first step the source cannot express — visible when View Native Query greys out.
Role-playing dimension
A single dimension table referenced by a fact table more than once, such as a date table serving both order date and ship date. Only one relationship can be active, so the others are created inactive and switched on inside individual measures.
Row-level security
Model-level filters defined as DAX expressions in roles, with users or groups assigned to those roles in the service. Because the filter is applied in the model, restricted rows never reach the client, unlike a report filter that only hides them.
Semantic model
The Power BI object that holds tables, relationships, calculations, and security definitions and answers queries from reports. It replaced the older term dataset and can be shared so many reports connect to one governed source of truth.
Semi-additive measure
A measure that can be summed across some dimensions but not across time, such as an inventory balance or an account balance. Point-in-time DAX functions like LASTDATE or the closing-balance family produce the correct value over a date range.
Sensitivity label
A Microsoft Purview Information Protection classification applied to Power BI content, which persists into supported exports such as Excel and PDF. It communicates confidentiality and can carry encryption, unlike endorsement, which communicates quality.
Star schema
A model design with a narrow fact table of measurable events surrounded by dimension tables of descriptive attributes, joined one-to-many. It compresses better and filters faster than a single flattened table carrying repeated attributes on every row.
USERELATIONSHIP
A DAX filter modifier that activates an inactive relationship for the duration of a single calculation. It is the standard way to handle a role-playing dimension such as a date table used for both order date and ship date.

Sources

  1. 1.Study guide for Exam PL-300: Microsoft Power BI Data AnalystMicrosoft Learn (accessed Aug 19, 2026)
  2. 2.Microsoft Certified: Power BI Data Analyst Associate — Exam PL-300Microsoft Learn (accessed Aug 19, 2026)
  3. 3.Microsoft Certifications — Exam frequently asked questionsMicrosoft Learn (accessed Aug 19, 2026)

Official sources

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

Last verified against the official exam content outline: