Another thought.

In datawarehousing, one often uses multiple cubes (fact tables) that not only use different combinations of your base dimensions but also different levels of summarization. For example, if your warehouse is based on a single star schema with sales being the main fact table, then you might have salesperson, customer and time dimensions. You might have many reports that don't care about the customer, so you create another cube, or fact table, that only has the salesperson and time dimensions. You might further care a lot about sales by quarter, so you may have yet another cube that has salesperson and time dimensions, but the time dimension is summarized to the quarter, rather than the day.

If you are doing a lot of grouping of a specific dimension (days rolled up to quarters), this might be a red flag for you to create a summarized cube. Now, I have yet to implement a star schema in Oracle, so I'm not positive about how to seamlessly implement these in Oracle. I would start with materialized views, however, as they would seem to do the trick.

HTH,

- Chris