- How sync and backup differ in purpose
- How one deletion reaches every device
- Four things sync cannot undo
- Generations, separate storage and restore tests
The short answer
Cloud sync does not replace a backup. Sync exists to keep every device in the same state, which means it faithfully replicates a deletion too. A backup exists to keep a copy of how things were at a particular moment. Different jobs. You need both before you can say you can get your files back.
How one deletion reaches every machine
The sequence matters more than the summary, so here it is step by step.
- Someone in sales tidies up the shared folder and drags a folder of old quotations to the recycle bin.
- The sync client on that laptop reports the deletion to the cloud.
- The folder disappears on the cloud side.
- The other five machines sync, and the folder disappears locally on each of them.
- The next morning, a colleague asks where the folder went.
Nothing malfunctioned. Every component did its job. Sync is fast, so by the time anyone notices, the folder is gone everywhere. Without a backup, your only remaining thread is whatever the provider still holds in its recycle bin or version history.
Recycle bins and version history are kept for a limited window, and the number of days differs by service and by plan. Notice three months later and you may be outside it. Check the retention period for the service you actually use, in your account settings or the provider's admin documentation.
What sync is not built to survive
Accidental deletion is only the most visible case.
- Overwrites. Even where previous versions are kept, both the count and the retention period have limits.
- A file saved while its contents were already corrupt. The broken state is treated as the current state and propagated.
- Ransomware that encrypts files. The rewritten files on the device are pushed up to the cloud like any other change.
- A sweeping change to permissions or sharing settings. The files are still there; the people who need them can no longer open them.
Japan's Information-technology Promotion Agency (IPA) makes a related point in its published material on ransomware: if an external hard drive or USB stick is left permanently connected to the computer, the backed-up files fall within the scope of the encryption too. Cloud storage that stays continuously connected through a sync client is closer to that situation than it looks.
Building the backup itself
Write down three decisions. If any of them is missing, there is a case you cannot recover from.
1. Keep generations
Copying only the current state somewhere else buys you nothing once a bad state overwrites it. Set a depth: seven daily, four weekly, six monthly, or whatever suits your data. The depth you choose is the grace period you get before someone notices a problem.
2. Keep it somewhere separate
Another drive in the same machine is not separate. Another folder in the same cloud account is not separate either. A single event — hardware failure, a compromised account — takes both at once. Use media that is not permanently attached, or a backup service that runs independently of your sync. IPA's material likewise advises preparing more than one backup device or medium, and connecting it only while the backup runs.
3. Test the restore
This is the step everyone skips. IPA's guidance is blunt about it: a backup you cannot restore from when the moment comes is pointless, so check periodically that a restore actually works. Once a year is enough. Pull one folder back to a scratch location and open the files.
Run the restore test while the person who set up the backup is on holiday. That is the only way to find out whether the written procedure is good enough for anyone else to follow.
How this looks in practice
Take a hypothetical example: a building-materials wholesaler with twenty staff keeps drawings and delivery notes in cloud storage, synced to everyone's machine. They added two things.
- A weekly copy to the office NAS, with the month-end copy moved to an external hard drive kept in the safe. The drive is connected only while that copy is made.
- A restore test every six months: last month's invoice folder is pulled back to a test location, and someone checks that the PDFs open and no pages are missing.
Six months later, a staff member dragged the invoice folder into the wrong place and could not find it. The cloud version history was enough to recover. The backup routine still earned its keep — knowing that the worst case was "back to the end of last month" let them search calmly instead of panicking. A lot of what a backup buys you is that margin.
Two cautions
Every backup copy is another place your data lives. If the files include personal information, decide how the media is handled and where it is stored before you start. A drive sitting on an open shelf is its own disclosure risk.
Also, a backup does not prevent the accident. It only lets you undo it. To reduce accidental deletions, you need different measures: restrict who can delete from shared folders, and list what is in scope before any clean-up session.
In summary
Sync keeps things identical. A backup keeps things from a moment in time. Once that distinction is clear, "it is in the cloud, so it is fine" stops being a reasonable position. Generations you can step back through, a location that is not permanently connected, and a restore you have actually performed — with all three, you have a backup.