I think this is an interesting question especially for those who would like to know how BI “thinks”.

The goal is to understand what rule is followed when several fact are at the same content level

Normally the server uses the content level to pick the correct fact table.

The server is looking for the most aggregated source.

First choice is a query in which the grain of the query matches the grain of the content filter.
If there are no sources which match that criteria, it will look to other means to chose.

It looks at how many dimensions are associated with the fact table (size of the content filter), levels of the content filter, number of values from the level definitions multiplied.