WordPress autoloaded options are settings loaded together so WordPress does not have to fetch each one separately. That is useful when the settings are needed. The problem is carrying a large collection of data that the current request never uses.
I built SlimPress to help review unused WordPress features, with a performance test that also reports autoloaded options. For this job, I would start with an inventory, not a database cleanup button. A large option needs an explanation before it needs deleting.
What the Site Health warning means
WordPress added an autoloaded-options check to Site Health in WordPress 6.6. The WordPress core explanation describes a warning when the combined size exceeds 800 KB. That is a reason to investigate, not proof that every option in the collection is unnecessary.
Open Tools → Site Health and look for the autoloaded-options result. Record what your own site reports before changing anything. I would not turn that warning into a promised speed improvement: the amount of unused data, the hosting and the requests you are measuring all matter.
Autoload is not inherently bad. WordPress's own guidance recommends it for options needed on every page because it avoids an additional database query. The aim is to stop loading unnecessary data, not to make the autoloaded total zero.
Start with a report, not a deletion
The SlimPress readme on WordPress.org describes a performance test that can report which autoloaded options each request loads from the database. That gives you something specific to investigate rather than treating the whole options table as disposable.
- Install and activate SlimPress. Nothing is switched off until you choose a switch.
- Open SlimPress under Settings and go to its performance-test area. Review the autoloaded-options report.
- Write down the options you want to investigate and the total your site reports. Keep that baseline outside the database you are about to work on.
- Work through the largest candidates first, checking who owns each setting and whether it is still needed.
You do not need to apply a preset or disable unrelated features to begin this review. I would leave the other switches alone so a database investigation does not become several different changes at once.
Large, unused and left over are different things
A large option may belong to an active plugin and contain settings that plugin still needs. Another may belong to something you removed. Size helps choose what to investigate first; it does not answer whether deletion is safe.
I would use an option's name as a lead, not as proof of ownership. Check the relevant plugin's documentation or source, or ask its support team what the option contains and how it should be maintained. If you cannot establish ownership, keep it on the investigation list instead of classifying it as rubbish.
For an option associated with a removed plugin, check whether you plan to reinstall it and whether another component still uses its data. Removing the plugin and deciding to discard its saved configuration are separate decisions. A backup is not a substitute for understanding that distinction.
What to do with a confirmed candidate
Take a restorable database backup before making database changes, and use a staging copy first where possible. Then choose the action with the plugin's own guidance in front of you.
- Still used: ask the plugin author whether the data needs autoloading and whether there is a supported way to reduce it.
- Confirmed obsolete: have a developer remove that specific option through an appropriate WordPress maintenance workflow, with its previous value backed up.
- Ownership uncertain: leave it alone until you can establish what reads or writes it.
Changing an option's autoload behaviour and deleting its value are not equivalent. The first changes when it loads; the second removes the saved data. I would not use a bulk operation that hides that difference.
What can break, and how to undo it
Deleting an option can remove configuration a plugin still needs. A lower total is not a successful result if the shop, forms or administrative workflow no longer behaves correctly. After an approved change, test the feature that owned the option as well as the pages you normally check.
Undo a database edit by restoring the affected data from the backup or using the owning plugin's documented recovery procedure. Turning SlimPress off will not bring back an option removed by another tool.
SlimPress's settings history is for its own settings changes. Its readme also warns that permanently deleted data is not restored by deactivation. Keep database recovery separate from undoing a runtime switch, and do not touch revision deletion or retention settings as part of this investigation.
When another approach is better
If you already have a reliable options report through your developer or maintenance tooling, use that rather than adding a plugin just to collect the same information. For a known problem in one plugin's stored data, that plugin's support team is a better next step than experimenting with unrelated WordPress switches.
SlimPress deliberately does not provide page caching or minification. An autoload investigation does not replace diagnosing slow queries, expensive plugin behaviour or the other work behind a slow page.
Record the baseline, investigate the largest candidates, change only what you understand, and verify the result. That is the approach I would take before calling any WordPress database cleanup finished.
Originally published on pluginlab.dev.
Top comments (0)