Checklists
Backup Restore Test Checklist
A practical checklist for testing website and database restores, so you know your backups work and how long recovery really takes before you need it.
- Published
- Updated
- Reading time
- 4 min read
A backup that has never been restored is an assumption, not a safety net. Backups fail for ordinary reasons: a database dump that silently stopped growing, files excluded by an old rule, credentials that no longer decrypt the archive, or a restore process that only one former employee understood. You only discover these problems by restoring.
This checklist walks through a complete restore test for a website or web application. It is designed to be repeated on a schedule, with each run documented, so that you build evidence over time that recovery works and know how long it takes.
Before the test: define what success means
A test without clear targets cannot pass or fail.
- Agree your recovery point objective (RPO): the maximum amount of data, measured in time, the business can afford to lose. The glossary explains RPO and RTO in more detail.
- Agree your recovery time objective (RTO): how quickly the site must be working again after an incident.
- List what must be recovered: website files, databases, uploaded media, configuration, environment variables, SSL certificates, DNS records and email if relevant.
- Identify the critical journeys you will use to prove the restored site works, such as forms, login and checkout.
- Decide the scenario. For example: full server loss, accidental database deletion, or recovery to a point before a malware infection.
If you are unsure how often backups should run to meet your RPO, read our guide on how often a website should be backed up.
Prepare an isolated restore environment
Never test a restore over your live site.
- Provision a separate server, container or staging environment with the same software versions as production.
- Prevent the restored copy from acting like the live site. Block it from search engines, disable outgoing email or route it to a test inbox, and switch payment gateways to test mode.
- Disable scheduled tasks that could send messages, sync stock or call external APIs from the test copy.
- Restrict access to the restore environment, because it contains real customer data.
Locate and retrieve the backup
This is where many tests reveal the first problem.
- Find the most recent backup using only your documentation, as you would in a real incident.
- Confirm it is stored off the production server, ideally with a separate provider or account.
- Check you can access the storage with credentials held in your vault, not in one person's memory.
- Decrypt the archive using the documented key or passphrase.
- Verify integrity using checksums where your backup tool provides them.
- Start a timer. Record when you began, so the total recovery time can be compared with the RTO.
Restore files and data
- Restore website files and compare the file count and total size with production.
- Import the database and check for errors during import.
- Compare record counts for key tables, such as orders, customers, posts or users, with production at the time of the backup.
- Check the newest records to confirm how recent the backup is. The gap between the latest record and the failure time is your actual data loss, which should be within RPO.
- Restore uploaded media, which is often stored separately or excluded by mistake.
- Apply configuration: environment variables, database credentials, cache settings and any secrets held outside the codebase.
Verify the restored site works
A restore is only successful if the site actually functions.
- Load the home page and key templates without errors or missing assets.
- Log in to the admin area with a test account.
- Run each critical journey you listed, such as submitting a form, searching, logging in and placing a test order.
- Check scheduled tasks and integrations start correctly once re-enabled in a controlled way.
- Review application and server error logs for warnings that appeared after the restore.
- Stop the timer when the site is fully usable, and record the total recovery time.
Record the results
Documentation turns a one-off test into evidence.
- Log the date, backup used, scenario and people involved.
- Record recovery time against RTO and data loss against RPO, and mark the test as passed or failed.
- Note every problem encountered, however small, such as a missing step or an outdated instruction.
- Update the restore runbook so the next person can follow it without guesswork.
- Assign actions for anything that failed, with an owner and a date.
Our article on why website backups often fail covers the most common problems these tests uncover and how to fix them.
Clean up safely
- Delete or securely retain the test environment according to your data protection policy. Restored copies contain personal data and must be protected or removed.
- Revoke temporary credentials created for the test.
- Confirm production backups are still running normally after the test.
Set the schedule
- Test restores at least quarterly for most business websites, and more often for online stores and applications where data changes constantly.
- Test after significant changes, such as a hosting migration, a new backup tool or a major platform upgrade.
- Vary the scenario between tests, including partial restores of a single table or folder.
- Include restore testing in your wider disaster recovery plan, so it sits alongside contacts, priorities and communication steps.
Want your restores verified for you?
Restore testing is built into SupportSafe maintenance plans and our website backup and recovery service. If you are not sure your backups would work today, start with a free website health check.