How do I optimize slow DAX measures in a Power BI report with millions of rows?

All answers to this question.

Optimizing large models usually starts with the Star Schema. Ensure you aren't using Snowflake schemas, as those extra joins kill performance. For your DAX, avoid using FILTER(TABLE, ...) and instead use KEEPFILTERS or filter specific columns like FILTER(ALL(Table[Column]), ...). This allows the Storage Engine to use its columnar compression and avoids the slow Row-by-Row processing of the Formula Engine. Also, check if you can pre-calculate the YOY values in a SQL view or during the ETL process in Power Query. Reducing the complexity of the "on-the-fly" calculation is the best way to ensure a snappy user experience for your stakeholders.

   Answered 2025-04-22 by Martha Stewart

Are you using any "Iterator" functions like SUMX or AVERAGEX over that 20-million-row table, or are the measures purely based on standard aggregations?

   Answered 2025-05-10 by Gregory White

• Gregory, I am using SUMX to calculate weighted averages across multiple dimensions. I suspect that's the bottleneck because it has to evaluate the expression for every single row before aggregating. I’m going to try moving that logic into a Calculated Column during the refresh period instead, even though I know it increases the model size. At least the report visuals will render faster for the end-users during their Monday morning meetings.

     Commented 2025-05-18 by Kimberly Thompson

Performance issues are often solved by "Auto Date/Time" settings. Turn that off in the global options and use a proper central Date Table to simplify your time intelligence.

   Answered 2025-06-05 by Justin Bradley

• Justin is spot on. Disabling Auto Date/Time reduced my model size by nearly 30% once, which immediately improved the calculation speed of all my time-based measures.

     Commented 2025-06-12 by Martha Stewart