Watch the video by Alex M Powers and adopt the techniques suggested to improve the efficiency (smaller size and faster model) of your Power BI solution! Adopt these ideas to establish best practices when importing data. This video was published by the Belgian Power BI User Group. Hint: you need to look at the cardinality of your data.

An example in the video shows splitting Date/Time into separate Date and Time fields lowers cardinality, reduces the number of unique rows in the table (behind the scenes), and therefore improves speed and efficiency.

It is also better to avoid using Calculated Columns and instead push back the calculation into the source (SQL Server, Excel, for example) because calculated columns are post compression. This means you will lose some efficiency if you add a calculated column.

Take aways:

  • Only bring in data you NEED
  • DateTime is BAD
  • Calculated Column is POST COMPRESSION

In the video he mentions using DAX Studio and the Vertipaq Analyzer.

Add a description of each field under Field Properties. It will display when user hover overs it.

Add synonyms for field names. If the user “Asks a Question”, they can use alternate field names you provide as part of the natural language (e.g. customer and “cust name”).

Add a Q&A button into the report to get users curious about natural language queries they can ask.

Slides can be found here.