Dataflow with Stale data

Have you ever combining tables and downstream you realized the joined fact table has multiple null value?

I wonder where else I could find a more complex diagram other than this one from @BIpolar Matthew Roches.

Such as F provides 2 output tables, both of which consumed by G at downstream, G was found with stale data occasionally in one of the output table, the refresh was successfully but data in one of the output is never refreshed.

Another case is E pass on data to F and X pass on data to Y; then E has output referred by Y; and F output is referred by X;

E and F in the same workspace; X and Y in another workspace; Then stale data occurred at Y which made me think the reference was treated as crisscross eventhough they are at different workspaces. Any suggestion?