Data inconsistency when using Mixed Mode

Hi.

I believe this this works as intended, as you usually want the fact-table to be filtered with your dimension tables. And you want your Direct Query to be as fast as possible. A best practice in this scenario is to always keep your dimension table accurate and up-to-date. A fix could be to create your dimension-table based on a SELECT DISTINCT from your fact_table regions.

In cases as this, it is possible to work around this issue by adding measures that looks up the region_dim, and region_description, and then use these in your table. This would affect performance and is not recommended if you got a big dataset (fact_table). Update your dimension-tables instead.

Work-around 

Create measures to look up the values:

Region dim = LOOKUPVALUE('Table3'[region];'Table3'[region],FIRSTNONBLANK('table2'[region],"")) 

Region description = LOOKUPVALUE('Table3'[region_description],'Table3'[region],FIRSTNONBLANK('table2'[region],""))

Create a table-visual to get the desired result:

Create a table-visual with the following columns and measures

Table2[sales_id],
Table2[sale_amount],
Table2[region],
[Region dim],
[Region description]

... and you should be good to go.