Yes
Makes the backups vastly leaner. Data is live data, archive data which can be reproduced, and data which is meaningless like work directories, transcoding dirs, etc. By segregating it, my backups take minutes of time as most data is not live data at all. And defining said backups doesn’t take exclusion lists, etc. Probably close to 50% of my so called data (space wise) is unnecessary to backup at all. Even nextcloud, it’s on my no backup dataset as I take backups from within the container per the nextcloud instructions on how to take a consistent backup. So, the backup is live data (and compresses well), but the actual data is not.