For years my homelab backups were a green tick in a cron log and nothing more. Borg ran nightly, the repo grew, and I felt safe. The trouble with feeling safe is that it costs nothing and proves nothing.
The thing that changed my mind was trying to restore a single Nextcloud config file after a botched upgrade. I knew it was in the backup. I just could not remember the passphrase prompt syntax, then the repo turned out to be on a disk I had quietly filled, so the last fortnight of snapshots had silently failed to write. The green tick was for the part that listed files, not the part that stored them.
So now I test. Once a month a second cron job picks a random snapshot, restores it into a throwaway directory, and checksums a handful of known files against the live copies. If anything drifts or the restore errors, I get an email that is impossible to ignore. It is not clever and it does not need to be.
borg extract --dry-run ::"$(borg list --last 1 --short)"
That dry-run alone would have caught the full disk. A backup you have never restored is a hope with good intentions. Restore one this week, even just to a temp folder, and find out which kind you have.