DistinctCount - aggregation - not working - I'm lost

Hi there!

I've been struggling with distinctcount per transaction_id when trying to use an agg.table.
It's been more than 7 months since this issue came to light, but I cannot see a solution, and I feel pretty clueless where i'm messing it up.

It works fine when in Directquery mode as long as I don't use an agg. table, but the results take up to 30~100 seconds to complete and is not a viable solution long term.

I set ut the sql table as an agg as well, so have access to both Powerbi, and SQL server.

The transaction_id is unique to each complete transaction, and we have around 50 million per year (it's around 16-17 digits long)
The fact table consists of transactions rows with a 1-10 (lines or products) per transaction_id

Agg table sql side:
...
agg_final as (
select
 store_number,
 product_id,
 cal_date,
 count(distinct(transaction_id)) as transaction_id
 sum(sales_amount) as sales_amount
 1 as fusion_key
count (*) as count_rows

from fact_sales

Group by  
store_number, 
product_id, 
cal_date
)

The dax i'm using is a simple calculate(distinctcount(transaction_id),TREATAS{1}, Fact_table[Fusion_key]))
The agg does get a match in dax studio, but the results are pretty wrong when using the agg table, Works perfect when I don't use agg. table.
Also using the product_name in a matrix.

The fusionkey does not change anything in the dax measure - seem irrelevant as GuyInaCube video tried to explain.
So it's either a sql side error or powerbi, unsure where to even debug.

Thanks for your time.