WordPress Auto Updates and What to Automate on a Fan Site
The fan wiki I used to run went down for eleven hours on a Sunday because a gallery plugin pushed a release that renamed one of its shortcodes. Nobody touched the site that weekend.

The fan wiki I used to run went down for eleven hours on a Sunday because a gallery plugin pushed a release that renamed one of its shortcodes. Nobody touched the site that weekend. The update arrived on its own at some point after midnight, the patch note pages that used that shortcode turned into walls of raw text, and I only found out when a reader messaged on Discord at eleven in the morning.
The lesson I took from it was not that auto updates are dangerous. It was that I had switched them on for all 31 plugins in one click, without asking which of those plugins would take the site with them if it went wrong. A volunteer run gaming site needs auto updates more than most, because the person who used to log in every Tuesday has exams or has quit the guild. The useful question is not whether to automate, it is which parts.
What WordPress Already Updates Without Being Asked
Most admins do not realise how much is automatic by default. Since version 3.7, core has installed its own minor and security releases, so a site on 6.8 will move itself to 6.8.1 and 6.8.2 without anyone logging in. Translation files update the same way. Major releases, the jump from 6.8 to 6.9, do not, unless you ask for them.
Plugins and themes are different. WordPress 5.5 added the per item auto update toggle you see in the Plugins list, and it is off for everything until you turn it on. That toggle is the one that broke my wiki, and it is the one worth thinking about carefully.
The mechanism matters because of how it fires. The check runs on WP-Cron, triggered by someone visiting the site, roughly twice a day. On a busy site that means updates land within hours of release. On a clan page that gets forty visits a day, mostly between eight and midnight, the update will install in the middle of your evening traffic rather than at four in the morning, which is the opposite of what you want. There is no built in setting for the hour. The only real control is moving cron to a server timer, which is a separate job.
Why Turning Everything On Is The Wrong Default
The common advice runs like this: enable auto updates everywhere, since an out of date plugin is the most common way a WordPress site gets compromised. The first half of that is true. Abandoned and unpatched plugins really are the usual entry point, far more often than weak passwords. The conclusion does not follow.
Sort your plugins by what happens when one breaks. A security scanner, an SMTP sender, an spam filter and a backup plugin can all update at three in the morning with no visible consequence. If one fails, your pages still render. Now take a page builder, a table plugin, a custom post type registrar or whatever renders your roster and your guide layouts. When one of those changes behaviour, every page on the site changes with it, and on a volunteer site that damage sits there until a reader complains.
Blanket automation treats those two groups as the same risk, and they are not. One group fails quietly, the other fails loudly and publicly. I have watched admins react to a bad auto update by switching the whole feature off, which leaves them with 30 unpatched plugins and a worse problem than the one they started with. The sensible split is roughly 80 percent automated and a short list left manual, and it takes about twenty minutes to set up once.
WordPress 6.3 did soften the risk. If a plugin auto update fails in a way that fatals the site, core now attempts a rollback to the previous version. That is real protection against a broken install. It does nothing about the case that actually hurts, which is an update that installs perfectly and then quietly changes how your content renders. No rollback triggers, because nothing crashed.
A Policy A Rotating Volunteer Team Can Actually Keep
This is the setup I use now on two gaming sites. The point is that it survives the admin disappearing for a month.
- Leave core minor updates alone. They are already automatic and they have an extremely good record. If someone disabled them with the
AUTOMATIC_UPDATER_DISABLEDconstant at some point, remove that line. - Decide on major core releases deliberately. Add
define( 'WP_AUTO_UPDATE_CORE', 'minor' );to wp-config.php to pin the behaviour so a future host change cannot alter it. Major releases then wait for a human, which is right for a site with a custom child theme. - Turn on auto updates for every plugin that does not touch the front end. Security, backups, mail, forms, analytics, image optimisation, spam filtering, SEO metadata. On a 25 plugin site that is usually 18 or 20 of them.
- Leave manual exactly the plugins that render content. Your builder, your table or roster plugin, anything registering custom post types or taxonomies, and any plugin whose shortcodes appear inside your guides. Write that list in a sticky draft post called Update Notes so the next admin inherits it.
- Check the update email. Core sends a message after every automatic install, to the admin address, from a wordpress@ sender that spam filters love to bury. Add a filter rule that labels it so it stops disappearing. If you have never seen one of these emails, that is the thing to fix first.
- Book one thirty minute slot a month for the manual list. Updates on a staging copy, click through the three page types you care most about, then run the same updates on production. Our walkthrough of a WordPress staging site before big updates covers the routine in detail.
Step four is the one people skip, and it is the whole value of the policy.
When An Auto Update Has Already Broken Something
Work out what changed before you touch anything. The Site Health screen has an admin event log under Tools, and your host's control panel will show file modification times, which together usually place the break within an hour. The update email, if you can find it, names the exact plugin and version.
Then roll back that one plugin rather than the whole site. A rollback plugin that installs a previous version from the official repository is faster and far less destructive than restoring a full backup over the top of three days of comments. Pin that plugin to manual afterwards and leave a note saying why, because the next admin will otherwise switch it back on.
If the failure is a blank screen rather than mangled layout, the cause is more likely a conflict than the update itself, and the process for isolating it is the same one described in our guide to finding a WordPress plugin conflict on a fan site. The other job worth doing in the same sitting is reducing the surface area. A site with 43 plugins has 43 chances of this happening, and most community sites can drop a third of them without losing anything, as the plugin audit for a clan site walks through.
One honest limit. I have never seen a core minor auto update break one of my sites in about nine years of running them, but that is maybe eight sites, which is not a sample anyone should treat as evidence. The reports that do exist usually trace back to a host with an unusual PHP configuration or a heavily modified theme, so if your setup is unusual, treat my confidence as mine rather than yours.
FAQ
Should I enable auto updates for themes?
For a parent theme you have not modified, yes. For a child theme there is nothing to update, and for a parent theme whose template files you edited directly, automatic updates will overwrite your work. If that describes your site, the fix is a child theme, not a disabled toggle.
Why did my plugin not auto update even though it is switched on?
Almost always WP-Cron. The check only runs when someone loads a page, so a quiet site can sit days behind. Load the site yourself, or check Tools then Site Health for a warning that scheduled events are missing, which points at cron rather than at the update system.
Do auto updates work if my licence has expired?
No, and this catches people out. Premium plugins check a licence key before serving an update, and an expired key means the toggle stays on while nothing installs. The dashboard rarely says so clearly, which is why an expiry date belongs in the same note as your manual update list.
Is it safe to automate everything on a small site with no traffic?
Safer than on a large one, because a breakage costs you little. The risk is that nobody notices for weeks. If you go that route, set an uptime monitor that checks a guide page rather than the homepage, since a mangled shortcode returns a perfectly healthy 200 response. More on keeping a small site stable sits in our WordPress section.
More in WordPress
WordPress
WordPress Custom Taxonomy for Game Patch Versions
The most common message our wiki got was never about strategy. It was some version of: is this still right after the update?
WordPress
WordPress Synced Patterns for Repeating Guide Blocks
Our clan wiki had a small box at the top of every guide: a line about which patch the guide was written for, a note that times...
WordPress
How to Fix WP-Cron When Scheduled Posts Miss on a Fan Site
You schedule the patch notes summary for 03:00, right when the servers come back from maintenance.