← All field notes

Should you restore WordPress files, the database, or both?

A theme edit broke the layout. A database mistake removed an article. These failures may need different recovery scopes, even if the backup tool offers the same prominent “Restore site” button for both. Reaching immediately for a full-site restore can undo valid work that happened after the backup.

Before restoring anything, identify what changed and what newer data you need to preserve. If the cause is unclear, take a protected copy of the current state for investigation and involve the person responsible for recovery. Do not experiment with destructive restores on the live installation.

Locate the failed change

WordPress stores content and settings in its database, while themes, plugins and uploads live in files. The WordPress backup handbook explains the two parts of a site backup. Recovery planning needs to account for their relationship, not just their separate locations.

A damaged theme file points you toward a file-level recovery or a tested redeployment. A deleted page may have a revision or trash recovery route before a database restore is necessary. An update that changed database structure needs closer attention because old files and new data may not be compatible.

Use the deployment history, error time and operator notes to narrow the scope. Do not assume that the most recent update caused every symptom if the site was already failing beforehand.

Ask what a restore would erase

List activity since the proposed recovery point: orders, form submissions, posts, user registrations and settings changes. Decide which of those must survive. On a busy site, that list can make a broad database rollback unacceptable without a more careful recovery plan.

For files, consider new uploads and legitimate code changes. A full file replacement can remove recent media even if the database still references it. A partial restore can create the opposite mismatch: old database records pointing at files that no longer exist.

  • Known code-only damage: assess a targeted file recovery and verify the dependent behavior.
  • Known content deletion: check application-level recovery options before replacing the database.
  • Coupled software and data change: test a coordinated recovery point in isolation.
  • Suspected compromise: preserve evidence and follow incident response, not just routine rollback.

Test the chosen scope away from production

Use an isolated installation to check whether the recovery point starts and whether the important workflows work. Record the versions, backup time and any repair needed. Keep email and payment integrations disconnected from real services during the exercise.

If the backup system uses an incremental chain, use its documented recovery process. Do not pick a small archive from the middle of the chain and assume it contains a complete site. Confirm which baseline and later parts are required.

Once the isolated result is understood, agree the production recovery window and the plan for preserving newer data. The right procedure depends on the application and the failure; a general blog post cannot safely choose database tables for an unfamiliar store.

Verify the business journey after recovery

Check more than whether the admin login works. Open recent content, inspect uploads and test the site’s main conversion path. Confirm that external integrations point to the correct environment and that scheduled work has not been duplicated.

Record the actual recovery point used and any known data gap. Tell the client what was recovered and what could not be recovered, without turning a successful homepage check into a claim of complete restoration.

A regular restore rehearsal makes these decisions less unfamiliar. The backup ownership worksheet helps identify who can carry out the recovery when the original hosting account is unavailable.


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