Test the restore, not the backup.
A business had backups running every night for years. When a drive failed, they found the backups had been writing to a folder that stopped syncing long before. The newest usable copy was very old.
A green tick in the backup software says a job ran. It does not say the data is complete, readable, or that you know how to put it back. You find that out the first time you try, and you want that first time to be a drill and not an emergency.
A restore test is small. It is also the only way to learn how long recovery really takes, which is the number you need when you decide how much downtime you can stand.
What to do
- Once a quarter, choose one real file at random and restore it to a different place. Open it and confirm it is the right one.
- Also restore one whole system or database to a spare machine or a spare folder. It does not need to be fast hardware.
- Time both from start to finish. Write the times down.
- Write the steps you followed as a short checklist, as if for a stranger. Keep a copy somewhere that is not on the machine being backed up.
- Fix anything that failed, then repeat the test on the next date you set.
- Make sure at least one copy lives somewhere separate from the original building.
When not to
Do not make this a heavy project for a laptop that holds nothing you cannot download again. Test what you would miss. If every important file lives in a sync service with version history, a simple yearly check that you can recover an old version may be enough.