filter dimension A by distinct count of dimension B per distinct dimension A

[cturner]

My data contains (among others) columns CLIENT and AGENCY.  I want to count the number of distinct values for CLIENT that have, per distinct value of CLIENT, distinct counts of AGENCY greater than or equal to two.  This seems like it should be straightforward, but I'm clearly missing something in the following DAX:

calculate(DISTINCTCOUNT('data'[CLIENT]),filter(all('data'[CLIENT]),DISTINCTCOUNT('data'[AGENCY])>=2)

Can someone kindly point out what I'm missing?

[mattbrice · Accepted solution]

Measure =
CALCULATE (
    DISTINCTCOUNT ( 'data'[CLIENT] ),
    FILTER (
        ALL ( 'data'[CLIENT] ),
        CALCULATE ( DISTINCTCOUNT ( 'data'[AGENCY] ) )>= 2 )
)

Give this a try.  

[cturner]

Thanks, that got it.  Didn't quite get the second calculate step, but it makes sense now.