Column order determines data visibility?

[Krenton]

Hi,

We have an application that comes with a database (so I can't change the design of the database).

I want to create a big dataset that people can use to build reports on.
For the dataset, I imported the tables separately (using queries) and linked them manually in Power Bi.

Below I created a little example with a part of the database to explain my issue.
I have a customer table, an address table, a contacts table, and a customer-contact-link table (a customer can have multiple contacts and the same contact can be linked to multiple customers/branches )

Now if I add a table to my Report view, with the fields Customer Name, City and Email, I get what I expect:

As you can see there is one Customer-City combination that doesn't have a contact email.

Now comes my issue,
If I exchange the position of the City and Email columns, I lose the row without Email:

 

I don't understand this behavior or how to fix this.
Maybe I think to much with a SQL background, but I would have expected this (modified screenshot):

Can someone please explain this to me?
I think I already tried every possible combination of cardinality and cross filter directions....

Thanks,
 

[lbendlin]

Please rethink your data model.  Bridge tables are no longer required, but having a distinction between dimension and fact tables will help you get a more deterministic behavior.  Avoid bidirectionals where possible.