The biggest win for me would be the ability to write efficient, custom-coded DAX that runs 100x faster than the SUMMARIZECOLUMNS queries that Power BI auto-generates.
Whenever I have a complex data model with many dimension and fact tables, I run into performance issues with SUMMARIZECOLUMNS. It has to do with the nature of that function and how it CROSSJOINS all the dimensions from different tables and must evaluate every measure for every combination in order to find the relevant rows to display. If I have three dimension tables with 1000 rows each, the DAX engine must evaluate 1000^3 or 1billion combinations for each measure. Yes, it's often pretty good at optimizing this, but the moment I have a measure that has any complex iterations, the DAX engine falls apart and just starts cranking through every combination taking the table visual minutes to load. Then, if I add a visual-filter to suppress certain rows, the execution time doubles as it executes the filtering query, then the original query.
I find that I'm often able to write my own query using SUMMARIZE that runs 100x faster. How cool would it be to just pop that DAX into the Table visual and enjoy amazing performance?
