microsoft pl-300 exam preparation and practice Advanced Practice Exam: Hard Questions 2025

You are building a Power BI model that imports a 1.2B-row fact table from a data lake (Parquet). Business users need daily refresh, but only the last 7 days change; historical data is append-only. During refresh, capacity CPU spikes and refresh time is inconsistent. You also must ensure that changed records in the last 7 days are correctly updated (not duplicated) even if late-arriving updates occur. Which approach best meets the requirements?

A dataset uses DirectQuery to an Azure SQL database. A report page has a slicer on Customer Segment and a matrix with Product Category by Month. Users report that selecting a segment intermittently returns 'visual has exceeded available resources' errors. You can make changes in Power BI but cannot change the database schema. Which change is most likely to reduce query load while preserving interactive slicing?

You ingest CSV files from multiple regions. The file structure is identical, but decimal separators vary by locale (some use comma, some use dot), and date formats vary (MM/DD/YYYY vs DD/MM/YYYY). You must build a single Power Query transformation that loads all files correctly and refreshes reliably in the Power BI service. Which approach is most robust?

A model contains a Sales fact table and a Date table. There are two date columns in Sales: OrderDate and ShipDate. Users need visuals where the x-axis is always the calendar from the Date table, but measures must switch between OrderDate and ShipDate based on a slicer selection. The solution must avoid ambiguous relationships and keep performance acceptable. What is the best modeling approach?

You are optimizing a star-schema model. A Dimension table (Products) has a unique ProductKey, but the fact table (Sales) contains occasional orphaned ProductKey values due to upstream data quality issues. The model is in Import mode. Users complain that totals look too high when filtering by Products because orphaned rows are excluded from product filters but still appear in grand totals depending on visuals. You want consistent, explainable behavior without altering the source system. What should you do?

A report has a measure: Total Sales = SUM(Sales[Amount]) A second measure is intended to show sales for the currently selected year on a line chart, but it returns the same value for every month: Sales YTD (Broken) = CALCULATE([Total Sales], DATESYTD('Date'[Date])) The Date table exists and is marked as a date table. However, the visual uses 'Date'[MonthName] (text) on the axis and 'Date'[Year] (whole number) in a slicer. What is the most likely root cause and best fix?

A single report page contains 18 visuals, each using different measures over the same fact table. Users complain that the page takes too long to render after any slicer change. You confirm that many visuals are independent and do not need to update together. Which design change is most effective to reduce perceived latency without changing the dataset storage mode?

You have a matrix showing Department (rows) by Month (columns) with a measure that calculates variance vs prior month. Some departments have missing months (no transactions). The variance must treat missing months as 0 sales, but only for the purpose of month-over-month variance (not for totals). Which DAX approach best meets the requirement?

A report uses a tooltip page that shows detailed metrics when hovering over points in a scatter chart. In production, the tooltip occasionally shows incorrect details that appear to come from a different data point. You find that the scatter uses a composite key (Customer + Product) but the tooltip page is filtered only by Customer. What is the best fix to ensure the tooltip always matches the hovered point?

Your organization uses Power BI deployment pipelines (Dev/Test/Prod). After deploying a dataset to Prod, scheduled refresh fails because the gateway mapping points to a non-production SQL server. You must ensure that future deployments automatically switch to the correct Prod data source without manual post-deploy edits, while still allowing Dev/Test to use their own sources. What should you implement?