Forum Discussion
Matrix multiple Sort columns
You may try using ISONORAFTER Function to add a measure.
Measure = VAR e = SELECTEDVALUE ( Table1[Employee] ) VAR c = [TotalCount] VAR t = SUMMARIZE ( ALLSELECTED ( Table1 ), Table1[Employee], Table1[Customer] ) RETURN COUNTROWS ( FILTER ( t, ISONORAFTER ( Table1[Employee], e, DESC, [TotalCount], c, ASC ) ) )
If you have the latest version of PowerBI desktop you can do this in 4 following steps:
Step 1: Convert the matrix into a table
Step 2: Sort the table by "Employee" column (Click again if you want to change the sorting order in opposite direction)
Step 3: Now add additional sort in the table by using Shift + Lift Click on "TotalCount " column (Shift + Lift Click again to change sort order to opposite direction)
Step 4: Convert the table back to matrix now, and your matrix retain these sort orders now
If this helps you then please give a thumbs up to the solution
This works well when the matrix only has one row.
But if it has two rows, aka hierarchy matrix, this doesn't work.
You can maintain multiple levels of the hierarchy by converting your Matrix to a Table, then removing your sub-level of the hierarchy from the table, do the multiple sort by steps in the table, convert the table back to a matrix, and finally add back in the sub-level of the hierarchy to the rows of the matrix. This should maintain the sort by order for multiple levels of the row hierarchy.
** Updated the steps for clarity (3/28/2025) **
Step 1: In your Matrix, add a field to the row hierarchy (which you will later remove).
Step 2: Change the Matrix to a Table visual.
Step 3: Sort all the columns you wish to sort, and in the order you wish them to be sorted.
Step 4: Delete the field you previously added in Step 1 from the table.
Step 5: Change the Table back to a Matrix visual.
Step 6: Ensure the right fields are back in the Row hierarchy.
This will maintain your multiple sorted fields within your matrix, even when you expand your row hierarchies.
Plase mark this as Solution. That is absolut genious and easier then a compolex measure 😉
this didn't work for me either
wow - thank you! this worked perfectly
