The short version
PL-300 has four domains. Three of them carry 25-30% each, and the fourth carries 15-20%. The weightings below come from Microsoft’s study guide, “Skills measured as of April 20, 2026”. The bullets illustrate how Microsoft assesses each skill; related topics may also appear. The traps and examples on this page are our own explanations, not Microsoft statements.
| Domain | Weighting |
|---|---|
| Prepare the data | 25-30% |
| Model the data | 25-30% |
| Visualize and analyze the data | 25-30% |
| Manage and secure Power BI | 15-20% |
Prepare the data (25-30%)
What it covers
Microsoft groups this into three parts. Get or connect to data: identify and connect to sources or a shared semantic model, change data source settings including credentials and privacy levels, choose between Direct Lake, DirectQuery, and Import, and create and modify parameters. Profile and clean the data: evaluate statistics and column properties, resolve inconsistencies, unexpected or null values, and import errors. Transform and load the data: pick data types, create and transform columns, group and aggregate, pivot, unpivot and transpose, convert semi-structured data to a table, create fact and dimension tables, merge and append queries, create keys for relationships, and configure loading.
In practice
A manager sends a spreadsheet where each month is its own column. You unpivot it so months become rows, set the types, remove blank rows, and split a combined product and region column. Then you split the data into a fact table and dimensions, so the model can be a star schema.
Typical traps
- Merge versus append. Merge adds columns by matching rows; append stacks rows. The join kind decides which rows survive, so read what the question wants to keep.
- Reference versus Duplicate. A reference depends on the original query; a duplicate is an independent copy. Questions ask about the impact of changing the original.
- Connection mode. The wording usually contains the clue: data volume, freshness needs, or whether the source supports DirectQuery.
- Privacy levels and credentials. They affect whether data from different sources can be combined, and they are set in the data source settings.
What to practice
Build a query chain with merges, appends, and a parameter. Break it with a bad data type and fix it. Look at column quality, distribution, and profile settings until you can say what each shows.
Model the data (25-30%)
What it covers
Design and implement a data model: table and column properties, role-playing dimensions, relationship cardinality and cross-filter direction, a common date table, and when to use calculated columns and tables. Create model calculations by using DAX: single aggregation measures, the CALCULATE function, time intelligence, basic statistical functions, semi-additive measures, quick measures, calculated tables and columns, and calculation groups. Optimize model performance: remove unnecessary rows and columns, find slow measures, relationships, and visuals with Performance Analyzer and DAX query view, and reduce granularity.
In practice
Your sales fact table has an order date and a ship date. Both relate to one date table, but only one relationship can be active. You keep order date active and use the other in a measure when you need ship date analysis. Sales per month now answers correctly for both views.
Typical traps
- Measure or calculated column. Measures respond to the filter context at query time. Calculated columns are computed at refresh and take space. Choose by what the question needs.
- Filter direction. Both-way filtering can fix a result and also create ambiguity or slowdown, so a question often asks for the smallest change that works.
- CALCULATE and the filter context. Predict what the filter does to existing filters before you pick an answer.
- Time intelligence. It needs a proper date table, and marking that table properly matters.
What to practice
Write the same measure three ways and compare the results. Use Performance Analyzer on a slow report page, then change the model and measure again.
Visualize and analyze the data (25-30%)
What it covers
Create reports: select and format visuals, build narrative visuals with Copilot, apply themes, conditional formatting, slicing and filtering, use Copilot to create or suggest report pages, configure the page, decide when a paginated report fits, and create visual calculations by using DAX. Enhance reports: bookmarks, custom tooltips, visual interactions, navigation, sorting, sync slicers, the Selection pane, drillthrough, export settings, mobile design, personalization, accessibility, and automatic page refresh. Identify patterns and trends: the Analyze feature, grouping, binning, clustering, AI visuals, reference lines, error bars, forecasting, anomaly detection, and Copilot summaries of the semantic model.
In practice
An executive wants a one-page view with a trend line and a ranked list. You choose a line chart and a bar chart, add a drillthrough page for the detail, and set tooltips so no one has to leave the page. You check that the tab order makes sense for keyboard users.
Typical traps
- Right visual for the question. Comparison, trend, composition, and distribution each have a natural visual.
- Filter versus highlight, drill down versus drillthrough. The names are close and the behavior differs.
- Selection pane orders. Layer order and tab order are separate settings.
- Report type. If the requirement is a pixel-perfect printable layout, think paginated.
What to practice
Rebuild one report using bookmarks, sync slicers, and a drillthrough button. Run it with the keyboard only. Add a forecast and an anomaly detection to a line chart.
Manage and secure Power BI (15-20%)
What it covers
Create and manage workspaces and assets: create and configure a workspace, configure and update an app, publish, import, or update items, create dashboards, choose a distribution method, configure subscriptions and data alerts, promote or certify content, identify when a gateway is required, and configure scheduled refresh. Secure and govern items: assign workspace roles, configure item-level access and access to semantic models, implement row-level security roles and group membership, and apply sensitivity labels.
In practice
You publish a sales report to a workspace, share it as an app for the regional managers, and use row-level security so each manager sees only their region. You schedule a refresh, and because the source is on-premises, you note that a gateway is needed.
Typical traps
- RLS gives rows, not access. A user in a role still needs access to the report or model.
- Roles versus sharing. Workspace roles are broad; item-level access is narrow. The best answer often depends on how much access a person really needs.
- Promote versus certify. Both are endorsement, with different meanings and who can apply them.
- When a gateway applies. It depends on where the source lives and how the model is connected.
What to practice
Create two RLS roles, test them with “View as”, and add a user. Then check which access that user still needs. Configure a scheduled refresh and read the refresh history.
More guides for PL-300
- PL-300 Exam Format, Scoring and Retakes
- Is PL-300 Hard? An Honest Difficulty Check
- PL-300 Prerequisites and Skills Check
- PL-300 Renewal and Retakes Explained
- Is PL-300 Worth It? An Honest Look
- PL-300 Cheat Sheet: Key Facts to Review
- DP-600 vs PL-300: Which Exam First?
Next step
Take the study plan and map these domains onto your calendar. To see these ideas in question form, try the sample questions, or test yourself on every domain with our PL-300 practice tests.