Some part of it can already be tested during creating the backup. It's not 100% foolproof of course and you should still test restoring once in a while but my backuping procedure goes kinda like this (simplifying for the sake of explaining):

    - compress backup
    - encrypt backup
    - add cryptographic checksum to encrypted backup file(s)
    - pass the newly created backup to a container that only has access in read-only to the volume with the backup
    - verify the cryptographic hash
    - decrypt
    - decompress
    - verify everything that can be verified
    - if all passes, whitelist the cryptographic hash for that backup as "known restorable backup"

So the very process of creating the backup already contains a lot of the checks needed to make sure you can restore the backup.

I still, once in a while, take a backup at random and verify that it can be restored but basically already by verifying that the cryptographic checksum matches, I'm pretty confident it's gonna work.

Never failed.

A proper backup procedure can be so much more than simply: compress/encrypt.

A real cycle, before whitelisting the backup as "probably restorable", is way more involved than that.