PL-300: Power BI Data Analyst — Exam Prep for Developers

Describe the PL-300 exam: four domains and their weights, time, pass mark, question formats, case studies and Microsoft Learn access during the exam

Connect to files, databases and the Aldervane CSV/JSON files, and recognise the M function each connector writes into the Source step

Explain where the data is at query time in Import, DirectQuery and Direct Lake, and how each one gets fresh data (refresh, live queries, framing)

Use column quality, column distribution and column profile to evaluate data, and switch profiling from the top 1 000 rows to the entire data set

Select appropriate data types in Power Query, including Fixed decimal number for money, Text for codes, and a correct conversion of yyyymmdd numbers to dates

Choose the right Power Query join kind (left/right/full outer, inner, left/right anti) for a requirement and predict the row count of a merge

Classify tables as facts or dimensions, state each fact table's grain, and recognise transaction, periodic snapshot and factless (bridge) facts

Choose the cardinality of a relationship (one-to-many, one-to-one, many-to-many) from the uniqueness of the columns and the grain of the tables

Explain why one shared (common) date table should filter every fact table, and what auto date/time does instead

Create single aggregation measures with SUM, AVERAGE, MIN, MAX, COUNTROWS, DISTINCTCOUNT and DIVIDE, and predict their value in any cell from the cell's filter context

Predict the result of a CALCULATE measure in any cell of a visual by stating the cell's filter context first

Write fiscal year-to-date, prior-year, period-shift and rolling-window measures with TOTALYTD/DATESYTD (year end "06-30"), SAMEPERIODLASTYEAR, DATEADD, PARALLELPERIOD and DATESINPERIOD, and predict their value in any month

Explain the measure explosion that calculation groups solve and when a calculation group is the right tool instead of separate measures or field parameters

Explain in VertiPaq terms why model size depends on the number of distinct values per column more than on the number of rows

Select an appropriate visual for a stated business question (comparison, trend, composition, distribution, relationship, single value, detail)

Explain what a visual calculation is, where it is stored, and what it can and cannot reference

State the requirements for Copilot in Power BI: paid Fabric F2+ or Premium P1+ capacity (no trial), Fabric Copilot capacity, the tenant setting, the cross-region setting and Q&A for page creation

Configure bookmarks with the Data, Display, Current page and Selected visuals options, and build a show/hide switch that does not reset the user's filters

Design a mobile layout for a report page and explain when the mobile apps show it

Use the Analyze feature (explain the increase or decrease, find where the distribution is different) and name where it does not run

Create and configure a workspace, including licence mode and contact list, and explain what a workspace is for

Decide for any data source whether a refresh in the Power BI service needs a gateway, and which kind: on-premises data gateway (standard mode), personal mode or virtual network data gateway

Assign the least-privileged workspace role (Admin, Member, Contributor, Viewer) for a given job, and know which roles row-level security restricts

Describe the PL-300 exam format that matters for planning: question count range, exam and seat time, question types, case studies, breaks and the 700 passing score

Save a Power BI Desktop file as a Power BI Project (PBIP) and name what each folder and file holds, including what Git should ignore

Translate WHERE, GROUP BY, HAVING, COUNT DISTINCT and CASE into DAX measures and DAX queries, and say who does the grouping in a report (the visual, not the measure)

Design dynamic row-level security from a security table (engagement team membership) instead of a single partner column, and predict what a given user sees

Place OneLake, lakehouse (Tables and Files), SQL analytics endpoint, warehouse, shortcuts and semantic models on one architecture picture, and say which engine writes and which reads