Prepare your data for use with Copilot for Power BI

• 4 minutes

Microsoft Power BI enables you to develop interactive reports within a single tool. Typical report development consists of the following steps:

• Prepare and model data
• Visualize and analyze data
• Secure and distribute reports

Each step can be time consuming and intimidating to new Power BI users, depending on the complexity of the data and the requirements for the report. You can reduce the amount of time spent creating reports by using Copilot for Power BI to perform certain tasks, such as:

• Create measures based on natural language.
• Update the semantic model with synonyms for improved user Q&A experience.
• Generate report content, summary visuals, and pages from prepopulated prompts.
• Analyze a summary of underlying semantic model.

However, you still perform the initial data cleansing and transformation tasks, which are crucial to ensure accuracy in reporting.

Ensure data quality

You also need to evaluate your semantic model for different aspects of data quality, or Copilot might not be effective.

Data quality is crucial when creating a Power BI report because it directly affects the accuracy and reliability of the insights you can derive from your data. Here are examples of how data quality can affect the successful creation of a Power BI report:

• Completeness: Missing values can cause gaps.
• Validity: Out-of-range data values can skew visuals and results.
• Consistency: Inconsistent data can affect date-related visuals.
• Uniqueness: Duplicates can affect data accuracy.
• Data Relationships: Cross-table visuals might not be possible without relationships.
• DAX Calculations: Limited calculations can result in fewer possible insights.

Prepare data with Power Query

Power Query is a key feature of Power BI Desktop to prepare your semantic model. It's the initial step in creating a Power BI report and is indispensable when using Copilot. Use Power Query to ensure data quality:

• Profile your data by assessing column quality, distribution, and profile.
• Clean your data by resolving inconsistencies, unexpected or null values, and other data quality concerns.
• Transform your data by implementing user-friendly naming conventions for columns and queries, altering column data types, and applying data shape transformations.