Should be straightforward - or perhaps I am missing some important point. Its hard to give a specific example as I have no information on the data structure in your model. If it is well structured for this analysis, it could be as simple as
Expression; Sum(DISTINCT SubGroup)
This sounds like a slowly changing dimension problem. Check out here:Google - Slowly Changing Dimensions and search this site for the same.
Thank you for your reply and suggested reading!
The underlying data can be categorized as a Type 2 Slowly Changing Dimension – so now I have a name for this malady!
Your suggested expression will indeed correctly count the number of unique subgroups within a filtered time period. Thank you.
The more perplexing issue is selecting the correct member from within the subgroup. David McMahon from QLIK will be helping me with this.