How to filter other dimensions in a star model (in a way that keeps performance good)

[GKJARC]

Hi all,

In Power BI Desktop I'm striving to create a proper star model. Where a FACTS table is used for the salesnumbers, and the Dimensions are used for additional info and filtering.
The issue I'm running into with a star model is that the unidirectional filtering makes slicers become unlogical to most end-users. 

Example: in a report, users want to make some selections on sales data. Slicing by Category 'Cars' for example.

Then they move on to the next slicer for further filtering by brand. However, the star model does not apply the Category-filter to any other data, only to the FACTS table. So even though they filtered on Cars, the report is still showing Motorcycle data as well.

I understand that it can be solved by changing the the relations to bidirectional, but that is not recommended in star models.

Another workaround I read on this forum is to create a measure COUNTROWS of FACTS_Sales, and then apply that measure as a filter everywhere in the report, showing only greater than 0. Adding this filter adds complexity to the report, plus it probably affects performance.

What are good way to deal with this?

[Anonymous]

you can set filter direction is BOTH in relation dim_product to fact_sales

[GKJARC]

Thanks very much @PeterSpillane  for your reply!

Assuming that a measure with INT ( NOT ISEMPTY (FACTS_Sales) is indeed the best way in terms of performance, I accept your reply as the solution. I do find it kind of weird though that Microsoft apparently expects users to put such measures as visual filters all over their reports in order to make slicers perform expected behavior.

[PeterSpillane · Accepted solution]

This post should give you exactly what you need.

https://www.sqlbi.com/articles/syncing-slicers-in-power-bi/