The backup you have never tried to restore.
The scenario:
Your backups run every night. The report shows a pass. Nobody has actually put one back.
A completed backup job is not the same as a working restore. Corrupted files, a folder excluded by mistake, an agent that stopped syncing three months ago and never threw an alert. You find out the gap exists at the exact moment you can least afford it, in front of a client who trusted you to have already checked.
The prompt:
You are building a backup restore testing process for a managed service provider.
Works in any AI tool.
Context:
- Systems backed up: [list]
- Current testing habit: [never / ad hoc / scheduled]
- Backup tool: [name]
Build:
- A restore test schedule set by system criticality, so the systems that matter most get tested most often
- The exact steps to run a test restore without touching the live environment
- A pass/fail checklist to sign off each test
- A log template that proves the testing happened, for a client or an auditor
- What counts as a failed test, and who gets told when one fails
A monthly test on your top 5 systems beats a quarterly test on everything. Start with the schedule.