Plan a restore test without touching the live site

Before you start

• Obtain the website owner’s approval and ask the responsible provider to confirm the supported test method.
• Choose a backup and list the functions you need to verify, using non-sensitive examples wherever possible.

What to do

1. Agree an isolated destination

   Record the test address and make clear that no live domain or database is being replaced. Ask the provider to control access to the restored copy. A noindex setting is not a password and is not a suitable way to protect customer records in a test environment.

2. Neutralise real-world actions first

   Before anyone opens a restored workflow, have the maintainer prevent real email, payments, bookings and external synchronisation. Document how those actions will be checked without reaching customers. Do not assume that a different website address makes integrations harmless or automatically changes them into test mode.

3. Check the agreed recovery targets

   Review representative pages, images, navigation and authorised editor access. Compare a known piece of content with the chosen backup date. Where a form or integration needs verification, use the agreed safe method and distinguish a simulated check from confirmed end-to-end delivery.

4. Record evidence and close the test

   Note what was recovered, what failed and any missing access or dependencies. Keep only the evidence you need and agree secure removal of unnecessary test copies with the provider. Update the recovery plan before relying on the same process during a real outage.

What to check

• The live website, domain and current records were not overwritten.
• The test did not send real messages or create real transactions.
• The result records limitations as well as successful checks.

Sources and further reading

Platform features can change. Check the provider’s current guidance before changing a live service.

• NCSC guidance on backing up and restoring data