WordPress auto-updates: on, off or selective?
Update notification. Busy afternoon. Live site. Do you click “Update now”, switch auto-updates off, or let everything install itself and hope?
No single setting is right for every site. It depends on what’s being updated, how critical the site is and whether someone can recover it if an extension misbehaves.
In my work managing client sites at Nivonix, updates are part of keeping things maintained, but they’re also changes to a live system. A sensible process balances security, compatibility and the team’s ability to test. “Automatic” isn’t the same as careless, and “manual” isn’t automatically safer.
Understand the update types
WordPress can apply some minor core updates automatically, and plugins and themes may have their own auto-update controls. What’s available depends on the version, configuration, extension and hosting. Check the settings on the actual site, because a guide written for another version may not match your dashboard.
Core, plugin and theme updates do different jobs. A core update changes the platform. A plugin update may change a form, payment integration, SEO output or security component. A theme update can alter layout and front-end behaviour. Treat them as different kinds of change, not one queue.
Some hosts also manage updates or backups. Find out what yours does and what’s still your responsibility. Two systems updating the same component cause confusion, while a managed update service may offer testing a small business can’t do alone.
Decide what can update automatically
For a low-risk, well-maintained plugin that isn’t business-critical, auto-updates can shorten the time it stays outdated. First confirm a current backup exists, the extension is supported and someone can check the site afterwards. Auto-updates without monitoring can simply hide failures until a visitor reports them.
For payment, membership, booking, multilingual or custom functionality, I prefer a more controlled process. Test in staging if you have it, read the release notes and update when someone can verify the site. If a staging environment isn’t affordable, prioritise a backup and a clear recovery plan.
The risk isn’t only that an update “breaks the site”. It can change a field, clash with another plugin, alter a template or stop a key form sending. A homepage can look fine while checkout or enquiries fail. Test the actions that matter to the business, not just whether the front page loads.
Build a maintenance rhythm
Keep an inventory of WordPress core, themes and plugins, with who uses each and what it does. Remove an extension only after confirming no feature or stored data depends on it. Record custom code and where it lives, especially if a theme update could overwrite it.
Schedule a regular review instead of waiting for an urgent warning. The rhythm should match the site’s risk, support model and update volume. A shop, a site with user accounts or one that publishes often needs more attention than a small brochure site. The point is a predictable owner and a repeatable check, not a magic weekly schedule.
Before a manual update, verify the latest backup and make sure restoring it has been tested at least once. A backup you can’t restore isn’t a recovery plan. For a significant change, note the current version, the time, what was applied and the results of post-update checks. It makes a later problem easier to isolate.
Use staging without treating it as a guarantee
A staging copy can reveal compatibility problems before they reach production. Keep it out of public indexing, and don’t copy sensitive customer data unless the team has an approved process. Test representative workflows and clear any test transactions or emails safely.
Staging can differ from live in PHP version, server configuration, caching or third-party connections. A passing test reduces uncertainty. It doesn’t remove it. Note the differences and still check the live site after deployment.
If an update is urgent for security, a long testing delay carries its own risk. Ask the host or maintainer what it fixes and whether a mitigation exists. Decide on the exposure, the compatibility information and your ability to recover, not on a rule to always wait or always update instantly.
What to check afterwards
Open the homepage, a representative post, a contact form and any core business flow. For a shop, check a product page, basket and checkout using the platform’s safe test process. Confirm notifications arrive, images load, navigation works and no unexpected error shows. Check logs or monitoring if you have them.
If something fails, note the symptom and the update that came before it. Don’t pile on more updates or change unrelated settings before you’ve narrowed down the cause. Restore from backup or roll back a component only through a process the site owner understands, because some updates include database changes that need careful handling.
Tell stakeholders what changed and what was tested. If something is still being investigated, say so. Clear communication matters, especially when the site brings in enquiries or sales.
Match update settings to recovery capacity
A useful policy says who receives update notices, who can approve a change and how the site is checked afterwards. If updates run automatically, monitoring should surface a failure quickly, and whoever’s on call should know how to pause or roll back the component. If updates are manual, define who reviews them and how long an exposed vulnerability can wait while testing is arranged.
Keep the policy easy for the site owner to follow. A list of exceptions with no explanation will eventually be ignored. Group components by role and risk, and review the groups when a plugin is replaced or a business function changes. The right settings are the ones your team can keep up consistently, with a recovery route you’ve actually tested.
Related reading
- WordPress staging sites: how to test changes safely
- Cleaning up a WordPress database without breaking anything
- A client onboarding checklist for website handovers
Look at the update settings for one important site and name who checks it after changes. What’s your recovery step if the next update breaks a key workflow?