Back, sorry. Funny I was discussing DAU concept last Friday. There is no direct way to calculate this, so we need to do a bit of MDX calculated measures.
I don't see another way as to iterate through all users, something like :
MEMBER [Next1] AS SUM( [Users].[Users Level],
IIF( isEmpty( [Measures].[MyEmpty] ),
0,
IIF( isEmpty( ([Time].currentmember.next,[Measures].[MyEmpty]) ),
0,
1
)
)
[Time].currentmember.next -> move to next day (assumes currentmember is day)
For each user we're checking if the users did connect today and if it's the case we check this for the next day.
The empty should be on a measure that quickly checks if the user was there for a given day. Not sure how a distinct count works on SSAS in this scenario (in icCube we don't care). A pity in icCube we could think extending aggregation to support vectors, so we could perform a faster version.