Sorry, I missed your reply. The problem with calculating the average of User Churn across all quarters is that User Churn is already aggregated and you can’t wrap an aggregated value in an avg.
In simple terms, your User Churn is basically lag(distinct count) divided by another distinct count which is fine. However, you can’t do avg(lag(distinct_count)/distinct_count). Normally to do this in QuickSight you have to replace the distinct_count inside the average by distinctCountOver but the lag function doesn’t work with distinctCountOver.
Is there any way you can calculate “Number of Users” and “Lost Users” at the dataset level? If you can do that, it should be pretty straightforward to calculate User Churn and the average of User Churn.