Maintenance
How to update WordPress without breaking anything
24 August 2026 · 6 min read
On this page
Updating is not risky. Updating without a way back, and without checking afterwards, is risky. This is the procedure we run on client sites, which takes about twenty minutes and almost never goes wrong.
For how often to run it, see the cadence guide. This is the mechanics.
Before you touch anything
1. Take a backup, now. Not last night's — one from the last five minutes. Files and database, stored somewhere off the server. This is your undo button, and its value is entirely determined by how recent it is.
2. Note what you are updating. Screenshot the updates screen, or write the list down. If something breaks in an hour, you need to know what changed. Four plugins updated and no record of which is a bad position.
3. Read the changelogs for anything major. You are looking for three words: "breaking", "removed", "requires". Anything jumping a major version — 5.x to 6.x — deserves thirty seconds of reading.
4. Pick a sensible time. Your quietest hour, and not a Friday afternoon. If you run a shop, not during a promotion.
Use staging if you have it
Most decent hosts include a one-click staging site free. If yours does, the safest version of this whole procedure is: clone to staging, update there, test, then either push or throw it away.
That turns a risky update into a rehearsal. It is the single biggest difference between how an agency updates a site and how most owners do.
If you do not have staging, the backup is your substitute — recovery rather than prevention, but workable.
The order
Core first, then plugins, then themes. This matters. Plugins are written to work with current core; updating a plugin against old core can fail in ways that updating core first would have avoided.
Within plugins, if you are updating several, do the big ones individually — page builders, WooCommerce, anything touching checkout — and batch the small ones. That way, if something breaks, you know which.
After every update
This is the part that gets skipped, and it is the part that matters.
- Load your homepage in a private window
- Load a key inner page — a service page, a product page
- Submit your contact form and confirm the email arrives
- If you have a shop, place a test order end to end, on mobile as well as desktop
- Log out and view the site as a visitor, since some breakages only affect logged-out users
- Check the browser console for red errors on the homepage
Five minutes. It converts "we updated" into "we know it still works", which are very different statements.
Step 3 catches the most expensive silent failure in WordPress. Email breaks quietly — a plugin update, a DNS change, a host tightening its rules — and you find out weeks later when someone asks why you never replied.
If something breaks
Roll back the single plugin rather than restoring the whole site. The full recovery steps are here, but the short version: deactivate the plugin you just updated, and if you cannot reach the admin, rename its folder over SFTP.
Restoring the entire backup should be the last resort, because it throws away everything since the backup — including orders.
Making it a habit
Same day, same time, every week. Sites drift out of date one deferred Tuesday at a time, and the drift compounds: the longer you wait, the riskier updating becomes.
If you would rather it simply happened, that is what our care plans do — this exact procedure, weekly, with a backup taken first, the site checked afterwards, and a rollback within minutes if anything moves. From $79 a month.