Impact of Incremental Dataset refresh partitions on the Power Query AddRankColumn function....

[MichaelP]

I have a table of records that contains date and time of record creations.

I wanted to rank records by date and time.

I used the Power Query AddRankColumn. 

It worked fine in PowerBI Desktop.

However I soon understood that I wasn't getting the expected ranking in PowerBI Service because I implemented Incremental Refresh on my dataset.

Basically what I observed was that records were Ranked within each partitions created by PowerBI Service.... Leading to completely useless and inaccurate my rank column.

Record Creation Date	Value	Partitions in Tabular Editor	Rank column value
June 14, 2023	We don't care	2023Q306	2
June 20, 2023	We don't care	2023Q306	1
July 3, 2023	We don't care	2023Q307	3
July 5, 2023	We don't care	2023Q307	2
July 21, 2023	We don't care	2023Q307	1
August 9, 2023	We don't care	2023Q30809	1
August 10, 2023	We don't care	2023Q30810	1
August 11, 2023	We don't care	2023Q30811	1

How would you suggest to ensure the ranking applies across partitions?

The reason I am doing that initiallly,  is that I wanted to filter the report with the last 2 or 3 extractions, so selecting rank 2 or 3, shall have addressed that.

NB: the source are files, auto extracted.

[MichaelP]

Update: I have actually used the DAX Rank function instead, which gives me what I want....

nevertheless, the partionning impact on the PowerQUery is still interesting to be aware off.