- CISO
- CIO
How Do You Prove a Backup Will Actually Restore?
THE SHORT ANSWER
A backup job that completes proves data was copied, not that the business can run again. To prove recovery, pick one critical service, restore it in an isolated environment from the copy you would use after an attack, and time it until a user can do real work. Compare that time with what the business expects.
On this page
Consider a company whose backup dashboard has shown green every night for a year. Then ransomware encrypts its main file and application servers.
The restore takes far longer than anyone expected. The backups are intact, but the identity servers have to come back first, the restore steps were written for an older setup, and nobody had agreed which systems mattered most. The backups worked. The recovery had never been tested.
What does a backup report actually prove?
A backup report shows that a job ran and data was written somewhere. It doesn't show that the data is clean, that the systems around it can be rebuilt, or how long the business will be down.
CISA's #StopRansomware Guide asks for more than a completed job: "Maintain offline, encrypted backups of critical data, and regularly test the availability and integrity of backups in a disaster recovery scenario." It explains why offline copies matter: many ransomware variants try to find and delete or encrypt any backups they can reach.
What do the standards say?
NIST's Cybersecurity Framework 2.0 sets the outcomes:
- PR.DS-11: "Backups of data are created, protected, maintained, and tested."
- RC.RP-03: "The integrity of backups and other restoration assets is verified before using them for restoration."
- RC.RP-05: "The integrity of restored assets is verified, systems and services are restored, and normal operating status is confirmed."
The framework "does not prescribe how outcomes should be achieved." A timed restore test is a practical way to show you meet those outcomes.
How do you run a timed restore test?
- Pick one service that matters, such as order processing or payroll, and write down what "working" means for it.
- Agree the target with the business owner: how long the service can be down, and how much recent data can be lost.
- Restore into an isolated environment, using the copy you would rely on after an attack, such as the offline or immutable copy.
- Start the clock at the decision to restore, and stop it when a user completes a real task in the restored service.
- Write down every dependency you hit, such as identity, DNS, licensing or network access.
- Compare the result with the target, fix the biggest gap, and repeat.
What does each kind of evidence prove?
| Evidence | What it proves | What it doesn't prove |
|---|---|---|
| Backup job success report | Data was copied | That the data is clean, or that it can be restored |
| File restore test | The backup tool can return individual files | That the service, and everything it depends on, can run again |
| Timed restore of a critical service | How long the business would actually be down | Recovery of every other service (each needs its own test) |
CISA also recommends keeping updated "golden images" of critical systems, so servers can be rebuilt quickly instead of being configured from scratch during an incident.
Why does this matter beyond IT?
The restore time is a business number. If leadership believes the company would be back in a day and the test shows a week, that gap affects contracts, customers and insurance. Cyber insurers increasingly ask for dated records showing that recovery worked, so the test result is also evidence at renewal.
Three checks you can run
| Check | What a pass looks like | What a fail looks like |
|---|---|---|
| When did you last restore a critical service end to end, and how long did it take? | A dated record from the last 12 months, with the time to a working service | No record, or only file-level restores |
| Is at least one backup copy offline or immutable, so an attacker with admin access can't delete it? | Yes, and the protection has been tested | Every backup can be reached with production admin accounts |
| Have business owners agreed how long each critical service can be down? | Written targets, signed off by the business | Targets assumed by IT |
Questions we hear
How often should we run a timed restore test?
CISA says to test "regularly" without setting a fixed interval. NIST's contingency planning guide uses annual testing in its sample policy. A sensible minimum is one timed test per critical service each year, plus a test after any major change to the systems or the backup setup.
Doesn't restoring a few files prove the backups work?
It proves the backup tool can return files. It doesn't show whether the service those files belong to can run again, or how long that takes. Most of the delay in a real recovery comes from the systems around the data, such as identity, networking and configuration.
Can we test without disrupting production?
Yes. Restore into an isolated test environment that can't reach production systems. That keeps the live business running and also stops anything restored, including any infection hidden in old backups, from spreading. Plan the network isolation before the test, not during it.
What is a golden image?
A golden image is a ready-built, up-to-date copy of a server or system configuration that can be deployed quickly. CISA recommends maintaining and regularly updating them for critical systems. During a recovery they save the time it would take to rebuild and configure each system by hand.
Sources
- CISA, #StopRansomware Guide (opens in a new tab)
- NIST Cybersecurity Framework 2.0 (February 26, 2024) (opens in a new tab)
- NIST SP 800-34 Rev. 1, Contingency Planning Guide (opens in a new tab)
- National Law Review, how cyber insurance carriers are raising the bar (September 29, 2026) (opens in a new tab)
