Get certified for free when you join Fabric Data Days 2026 and dive into Fabric, Power BI, SQL, AI, and other essential data skills.
Join nowData Days is here! Join us now for 60+ days of learning, challenges, and connection. Learn more
Hello,
I am working on a model that, simplified, looks like the one below. The order table (fact table) is linked to a calendar through 2 relationships:
Create date and calendar date - active
Complete date and calendar date - inactive
The tables look like the one below:
Calendar_table
Order_table
Complete blank can be blank if the order is not complete yet.
I need to calculate the sum of units for a specific period of time for only the complete orders. The measure I created looks like this:
Units = CALCULATE(SUM(Order_table[Units]),
USERELATIONSHIP(Order_table[Complete Date], Calendar_Table[Calendar Date]), //to make the relationship active
FILTER(ALL(Calendar_table), Calendar_table[Calendar Date] >= period_start && Calendar_table[Calendar Date]<=period_end))
I am not sure why, the only orders I get are the ones without any dates (the incomplete ones). What am I doing wrong?
I appreciate your help!
Solved! Go to Solution.
Hi @Anonymous ,
It looks like you're trying to combine two different techniques into a single measure.
You usually create a measure as either
-1- A DISCONNECTED measure (i.e. No relationships) and use FILTER in the way you have in order to identify the rows you want to use...
-OR-
-2- You use a CONNECTED measure, whereby you identify the rows you want to use via an active relationship or via USERELATIONSHIP with an inactive relationship.
Personally, I tend to make ALL relationships INACTIVE when I have multiple of them to the same table (usually Calendar to Fact[Created]/Fact[Completed], exactly how you're doing here) so I have complete control over when the relationships come into play (via USERELATIONSHIP).
If both of your relationships were inactive, this measure should work:
_completedUnitsByPeriod =
VAR __periodStart = //however you want the user to select
VAR __periodEnd = //however you want the user to select
RETURN
CALCULATE(
SUM(Order_table[Units]),
FILTER(
Order_table,
Order_table[Complete Date] <= __periodEnd
&& Order_table[Complete Date] >= __periodStart
&& NOT ISBLANK(Order_table[Complete Date])
)
)
However, in your current scenario, the following should work fine with the relationship making the row selection:
_completedUnitsByPeriod =
CALCULATE(
SUM(Order_table[Units]),
USERELATIONSHIP(Calendar_table[Calendar Date], Order_table[Complete Date])
)
I may have misunderstood exactly what you want your measure to output, but the principle is sound.
Pete
Proud to be a Datanaut!
The first option worked for me. Thank you!
Hi @Anonymous ,
It looks like you're trying to combine two different techniques into a single measure.
You usually create a measure as either
-1- A DISCONNECTED measure (i.e. No relationships) and use FILTER in the way you have in order to identify the rows you want to use...
-OR-
-2- You use a CONNECTED measure, whereby you identify the rows you want to use via an active relationship or via USERELATIONSHIP with an inactive relationship.
Personally, I tend to make ALL relationships INACTIVE when I have multiple of them to the same table (usually Calendar to Fact[Created]/Fact[Completed], exactly how you're doing here) so I have complete control over when the relationships come into play (via USERELATIONSHIP).
If both of your relationships were inactive, this measure should work:
_completedUnitsByPeriod =
VAR __periodStart = //however you want the user to select
VAR __periodEnd = //however you want the user to select
RETURN
CALCULATE(
SUM(Order_table[Units]),
FILTER(
Order_table,
Order_table[Complete Date] <= __periodEnd
&& Order_table[Complete Date] >= __periodStart
&& NOT ISBLANK(Order_table[Complete Date])
)
)
However, in your current scenario, the following should work fine with the relationship making the row selection:
_completedUnitsByPeriod =
CALCULATE(
SUM(Order_table[Units]),
USERELATIONSHIP(Calendar_table[Calendar Date], Order_table[Complete Date])
)
I may have misunderstood exactly what you want your measure to output, but the principle is sound.
Pete
Proud to be a Datanaut!
Don't miss out on Data Days, June 15 through August 7. Learn Fabric, Power BI, SQL, AI and more.
Check out the May 2026 Power BI update to learn about new features.
| User | Count |
|---|---|
| 4 | |
| 4 | |
| 3 | |
| 2 | |
| 1 |
| User | Count |
|---|---|
| 11 | |
| 11 | |
| 4 | |
| 4 | |
| 4 |