← All field notes

How to test a WordPress backup before you need it

The backup job finished last night. Good. Now answer a harder question: could someone other than the person who configured it restore the site with the access your agency currently holds?

A restore exercise tests that question without an outage hanging over it. You are checking the recovery process as much as the archive. Missing bucket credentials, an undocumented encryption key or a host-specific assumption can stop a recovery before WordPress even starts.

Choose a safe destination

Use a disposable installation that cannot send real customer email, charge payment methods or publish content to external services. Restrict access to the test environment. A search-engine visibility setting is not access control, especially when the copy contains customer data.

Write down what you intend to recover and which recovery point you selected. For a busy site, also note what data arrived after that point. You need to understand the gap before deciding that a particular backup is an acceptable emergency fallback.

Never use the live site as the destination of a practice restore. If your tool requires destructive confirmation, stop and verify the hostname, database and storage path together. A familiar logo in the admin bar is not proof that you are in the right installation.

Run the recovery with the documented access

  1. Retrieve the backup from its actual offsite destination, not a convenient local copy.
  2. Follow the recovery instructions using the credentials the fallback operator would have.
  3. Restore the files and database required by the selected backup method.
  4. Check the restored URL and environment configuration before opening integrations.
  5. Record any manual repair, missing dependency or undocumented step.

The WordPress backup lesson explains that a full site recovery needs both files and database data. Your rehearsal should demonstrate that the two belong together. An archive full of uploads cannot reconstruct the settings and content stored in the database.

Inspect more than the homepage

Open a recent article and its images. Sign in with an appropriate test account. Check a permalink, a search result and the main navigation. If the site takes enquiries, submit a test message into a controlled inbox and confirm the expected fields arrived.

For a shop, inspect product variations and a representative order record. Use sandbox payment settings if you test checkout. Do not reconnect live webhooks merely to make a test pass; isolate them or agree a controlled procedure with the service owner.

Compare a few known files or records with the recovery point you expected. You are looking for silent gaps: missing uploads, a stale database, broken paths or configuration that only worked on the original host.

Leave a recovery note, not just a pass mark

Record the time spent obtaining access, transferring data, restoring and verifying. These are measurements of this exercise, not a promise that every future incident will take the same time. A host outage or a much larger database can change the result.

Name the backup, destination, operator and checks completed. Add the problems found and who will fix them. Remove the disposable copy according to your data-handling policy when the exercise is finished; retain the evidence, not an unnecessary duplicate of customer information.

Repeat the test when the recovery method changes, when the site grows substantially or when the person responsible leaves. Put the next exercise into your maintenance checklist. Before closing the exercise, ask the fallback operator which instruction they had to ask you to explain. Fix that instruction while the difficulty is still fresh.


Have a correction or a workflow we should cover? Tell the IrmaWP team.