Hi.
I’m having a conceptual problem with tracking sales vs. call center calls.
Each record in the fact table represents a call to the call center.
In this record are various facts and foreign keys that map to marketing campaigns, etc.
The product sold ID is NULL on no sale, and filled in for a sale.
When creating MDX to retrieve data by campaign, for example, to track the number of calls for a campaign vs. it’s sales, that works fine (because there’s a column labeled sale that’s either 0 or 1 for the sale and I just sum it). This way I get the conversion percentage of calls to sales.

But when creating MDX to track the product sold, as soon as I do a crossjoin on campaign and products sold, for example, I lose the total calls- the number of calls is the same as the sales (similar results as a SQL join). Am I doing something wrong or is it conceptually impossible to track this type of metric down to the product when there are NULLs in the fact table? I’ve tried converting the nulls to a NONE category, but that doesn’t stop the crossjoin from not giving me the desired results, plus I then have to filter out NONE as a product.
Is that why a lot of sample warehouses have a separate sales cube?

Thank you,
Richard