Test whether a backup can actually restore your files
Choose representative files, restore them to a separate place and verify content and access.
Follow the method ↓
The answer in 30 seconds
A green backup status does not show whether the right files were included or whether you can recover them after losing a device. A small restoration test provides evidence without risking the working copy.
Select a few important files of different types, restore them into a new folder, open them and document the time, version and account needed for recovery.
Examples to adapt
Restore one older and one recent photo and check resolution and date.
Open a restored version and confirm it is not an empty placeholder.
Check whether account recovery depends on the phone that was lost.
Follow the method
- 1Choose samples
Pick files from different folders and dates, including one you know changed recently.
- 2Restore separately
Use a new folder or spare device; do not overwrite your only good copy.
- 3Open and compare
Verify content, file size, version and any application needed to read it.
- 4Record gaps
Note missing folders, recovery credentials and the time a full restore might take.
A closer look
Separate a completed copy from a successful recovery
Fictional scenario: a folder contains a text document, a spreadsheet and a photo. Backup software reports completion. A separate recovery test provides additional observations.
| Observation | Supported conclusion | Next check |
|---|---|---|
| The task is marked complete | The software reports execution | Check the copy’s date and contents |
| All three files are recovered elsewhere | These files can be recovered under the test conditions | Open them and check their contents |
| Text opens but the spreadsheet is unreadable | The test is only partly successful | Check required versions and software |
| Date and outcome are recorded | This test can be compared with the next one | Repeat after a significant change |
The documentation recommends testing backups. A sample test supports conclusions only about the files actually recovered and opened.
Where should the recovered test files go?
Use a separate location and retain the originals. Observe recovery without overwriting current work. Record the tested copy’s date and any required software or passwords, without putting secrets into a shared record.
Supporting documentation : Cybermalveillance.gouv.fr — Sauvegardes · Consulted on 2026-10-15. Examples and calculations are illustrative.
A checklist to keep
Use these checks to record what you found. The grid supports a decision; it does not make one for you.
| Check | What to examine | Action |
|---|---|---|
| Sample | Does it represent important data? | Choose |
| Destination | Is it separate from the original? | Set |
| Version | Is it the expected date? | Compare |
| Access | Can the file be opened? | Test |
| Gap | What is missing? | Record |
Read the checklist directly in the table.
What to check
A synced folder may mirror a deletion rather than provide recovery.
An untested encrypted backup can fail if its key is unavailable.
A single successful file does not prove every folder is covered.
Common questions
How often should I test?
Repeat after major device or backup changes and at an interval matched to the importance of your data.
Should I overwrite current files during a test?
No. Restore a sample separately so the test cannot destroy the working copy.
Sources and documentation
Documentation consulted on . Examples are illustrative; interfaces and results may change.
Apply this method to a wider task
Restoring a backup: which files should you retain?
The newest file is not always enough to restore your data. The backup type determines which components recovery needs.
- Identify the backup set
- Locate dependencies
- Open a restored copy
| Situation | What changes | Useful check |
|---|---|---|
| Full | Copies all selected data. | Check the scope and excluded folders. |
| Incremental | Changes since the previous backup. | Retain the starting full backup and required increments in order. |
| Differential | Changes since the latest full backup. | Retain the starting full backup and chosen differential. |
Test a non-sensitive file in a separate folder without overwriting the original. Record the recovered version and errors. Disconnect the medium after use: a connected copy can also be affected by an incident.
Documentation : Cybermalveillance.gouv.fr — Sauvegardes ↗ · · Read the reference record →