cross process(data marts) ratios

We have 8 data marts in production and we are now designing Enterprise Data Warehouse(EDW). We will now like to identify & develop second line data marts, ex. profitability dara mart created from Revenue and Cost star schemas. This useful cross process metric may be derived from multiple fact tables either from same subject area or different subject area (say Finance & Inventory).

What are the industry best practices/guidelines in accomplishing this?
- when to develop them?
- when to create physical tables vs drill across?
- where to store them - separate schema?
- ETL issues to watch

Hope you will share your experience & thoughts on this.

Thanks