Been working a lot with Azure Data Warehouse, so knew this one pretty easy.
I surely choose ROUND_ROBIN a lot for a number of tables where I just don't have a good candidate for a distribution key. This is where I can't find a field of data that can evenly be distributed across all 60 databases behind the scenes where I can minimize data movement and reduce the skew.
There are some interesting ways to get around default values though. Having a lot of fun exploring the methodologies here.
Statistics are your friend if you op for ROUND_ROBIN.