RANKX() over multiple columns?

[gjohnson900]

I've got this as incoming data...

I'm sending the data to a Sankey chart, where the "weight" is the DISTINCTCOUNT of customerkeys.  I want to be able to offer the user a Top N chiclet slicer to keep the chart readable.  To that end, I created a computed measure [dRanker], that will RANKX the number of customers over my [source] values...

The problem is that the Sankey chart needs both a [source] and a [dest] to make it useful.  When I add a visual filter on the Sankey to only show where dRanker is <= 2, I get 9 different [source] values, not just the 2 highest ranked...

When I add [dest] to my data table, I can see that many more [source]s are getting RANKed as <= 2...  

I'm guessing that's why I'm getting so many results in the Sankey as well, so...  How do I tell RANKX() that I want it to rank over the DISTINCTCOUNT of CustomerKeys for BOTH [source] and [dest]?  I feel like the tSQL RANK() function allows this, but I can't seem to find similar overload functionality for RANKX() in DAX.

Another approach that would be just as good...  How could I get the Sankey chart to only show [source] values that appear in the original table above where the [source] ranking is correct?

Here's the dRanker DAX...

dRanker = RANKX(ALLSELECTED(FW17[source]), [CustCount])

... and the [CustCount] calculated measure...

CustCount = DISTINCTCOUNT(FW17[customerkey])

Am I thinking about this wrong?  What's wrong with my brain.  It's usually pretty good.

[v-juanli-msft]

Hi @gjohnson900

First you need to detemine how to calcuate the DISTINCTCOUNT of customerkeys, in your previous step, you just add the column [customerkeys] to the table  and select "count(distinct)", right? in this way, the result will change with the columns you add to the table.

Thus your [Rank] column also become different.

So first please let me know how you would like the DISTINCTCOUNT of customerkeys.

Best regards

maggie