Data Model Cross-Filtering between 2+ Tables

[raymond]

Hello everyone,

I guess it is a seemingly simple problem but I dont know how to do it. There are 3 tables that I want to cross-filter. At the end I want to set a filter on which I can select e.g. (1) "Channel » Google_1" and get the number of Events from the Conversion Table that contribute to that. In another case I want to set a filter with (2) "Conversion » Download" and get the number of clicks and impressions. 

My first approach to solve that was to use the 4th table as some kind of a matching table. This however only worked for (1) but not for the reverse filter setting (2). 

Does anyone have a clou? 

A sample PBI file can be found here: Sample File

A Channel Table 

Channel	Impressions	Target
Google_1	1000	Produkt
Google_2	50	Start
Xing_1	2000	Produkt
Xing_2	250	Start

A Content Table

Page	Clicks	Source Channel
Produkt	100	Google_1
Start	5	Google_2
Produkt	200	Xing_1
Start	2	Xing_2

A Conversion Table 

Conversion	Events	Source Channel	Source Page
Download	5	Google_1	Produkt
Download	1	Google_2	Start
Download	10	Xing_1	Produkt
Download	1	Xing_2	Start
Contact	5	Google_1	Produkt
Contact	0	Google_2	Start
Contact	10	Xing_1	Produkt
Contact	0	Xing_2	Start

[Kristjan76]

You should be able to keep just the three source tables, and then make a relation from the channel table to the other two tables. And that is more or less it, if this solves the problem mark it as solved.

[raymond]

Thanks for your help. Unfortunately it doesnt resolve the isse.

1. There is a warning due to the many to many relationship which I would like to avoid.

2. When I filter by Channel (e.g. Google_1) the associated clicks and events dont get filtered automatically and vice versa.

I was thinking of including matching tables but this doesnt work right as well. It does however show me the right values when I start from the channel table. The further I go, the more inconclusive the values gonne get. The Content table only shows two columns correct. The Conversion table only shows conversions correct.