How Often Do WordPress Updates Break Sites? Stats Traced
How often do WordPress plugin updates break sites? We traced each popular statistic to its source. Most measure worry, not a failure rate.
If you search for how often WordPress updates break sites, you will find confident percentages. We opened the source behind each one we could find. None of them measures what most people think it measures, and one could not be traced at all.
Below: what each number actually counted, who published it, when, and whether we could confirm it on the original page.
What percentage of WordPress plugin updates break sites?
We could not find a published, checkable figure for the share of plugin updates that break a site. The numbers people quote are survey answers about worries or past experiences, not failure rates per update. Anyone quoting a precise “X% of updates break sites” figure should be asked for the source.
That is not the same as “updates rarely break sites”. It means nobody has published the data in a form you can check. Vendors that test updates automatically probably hold it, but the pages we read don’t share it.
What is the difference between a per-update failure rate and per-site breakage?
These are three different measurements, and most articles mix them up.
| Measurement | What it divides | Example question it answers |
|---|---|---|
| Per-update failure rate | Updates that broke something ÷ all updates applied | “If I click Update on this plugin, how likely is a problem?” |
| Per-site breakage over a period | Sites that broke at least once ÷ all sites, over a year | “How likely is it that my site breaks at some point this year?” |
| Pain-point share | Survey respondents who name updates as a worry ÷ all respondents | “How many people are worried about this?” |
The first two can tell very different stories. Here is made-up arithmetic, only to show the effect: a site that applies 200 plugin updates in a year and has one bad one has a per-update failure rate of 0.5%. The same site counts as “broken this year” in a per-site figure. A survey that asks “did an update break your site in the past year?” measures the second kind, and with enough plugins and enough updates, a high yes rate is expected even if each single update is very likely to be fine.
The third kind measures fear, which is real and worth knowing, but it is not a rate of anything breaking.
Where do the popular WordPress update statistics come from?
| Figure as usually quoted | Published by | Date | What it actually measured | Confirmed on the original page? |
|---|---|---|---|---|
| “35% of WordPress users struggle with site-breaking updates” | Pagely (managed WordPress host), blog post by Lizzie Kardon | 1 May 2018 | Share of respondents who picked “site-breaking updates” in a multiple-choice question about their biggest WordPress headaches | Yes. Sample size is not stated on the page |
| “52% of WordPress professionals had their site break from a plugin or theme update in the past year” | Usecue (a web development studio’s blog) | 1 October 2025 | Attributed to “a 2023 WP survey”, with no publisher, link or sample size | No. We could not find the survey it refers to |
| “WPBeginner estimates hundreds of thousands of sites break after every major release” | Attributed to WPBeginner by the same Usecue post | 2025 (Usecue) | The only matching wording we found is a reader comment on WPBeginner’s WordPress 6.6 preview, about the new rollback feature (see below) | No. WPBeginner’s own text does not say it |
| “The biggest risk when updating plugins is a fatal error” | Make WordPress Core, Rollback Auto-Update merge proposal | 19 April 2024 | The reasoning for a new WordPress feature. It gives no frequency | Yes. No numbers on the page |
| “91% of new vulnerabilities are in plugins” | Patchstack, State of WordPress Security in 2026 | 2026, covering 2025 | Security vulnerabilities found, not updates that broke sites | Yes. Relevant to security, not to breakage |
The Pagely 2018 survey: 35% named site-breaking updates
This is the most solid of the lot, and it is often misquoted. Pagely asked WordPress users, from agencies and bloggers to enterprise tech leads, about their biggest headaches. In a multiple-choice question, the top answers were performance issues (52%), security issues (41%) and site-breaking updates (35%) (Pagely).
So the accurate sentence is: “In Pagely’s 2018 survey, 35% of WordPress users listed site-breaking updates among their biggest headaches.” It is not “35% of updates break sites”, and it is not “35% of sites broke”. The page does not say how many people answered, and the survey is now more than eight years old. WordPress has added update rollback features since (more on that below).
The Usecue claim: 52%, source unknown
Usecue’s post “Help, my website is down!” (1 October 2025) says a 2023 WP survey found that 52% of WordPress professionals had a site break because of a plugin or theme update in the past year (Usecue). It names no publisher and links to nothing. We searched and could not find a 2023 survey with that result.
One thing we noticed but cannot explain: the 52% figure matches Pagely’s 2018 share for performance issues, a different question in a different year. We can’t tell whether that is the origin or a coincidence, so treat the 52% as unsourced. The post is also an argument for moving away from WordPress to static site generators, which is worth knowing when you weigh it.
The WPBeginner “hundreds of thousands” line
The same Usecue post says WPBeginner estimates that hundreds of thousands of sites break, or show visible errors, after every major release. The WPBeginner article we found is its preview of WordPress 6.6 (17 June 2024). Its own text says WordPress 6.6 will add automatic rollback for plugin updates that break a site (WPBeginner). The closest wording is a reader comment under that article, saying the feature “will save hundreds of thousands of sleepless nights”.
That is a figure of speech from a commenter about a new feature, not an estimate of how many sites break. We did not find a separate WPBeginner estimate.
The WordPress core team: a real risk, no number
The best evidence that update breakage is common enough to matter comes from WordPress itself, and it has no percentage attached. The Rollback Auto-Update merge proposal calls a PHP fatal error that crashes the site the biggest risk a site owner faces when updating plugins (Make WordPress Core). The feature it proposed was aimed at WordPress 6.6: during an automatic plugin update, WordPress loads the homepage, and if it finds a fatal error, it puts the previous version back.
Earlier, WordPress 6.3 added a rollback for manual plugin and theme updates where the update process itself fails, for example an empty download or files that can’t be moved into place (Make WordPress Core).
Both are narrow on purpose. Neither catches an update that installs cleanly, leaves the homepage working and breaks the checkout or a form.
Security statistics are not breakage statistics
Patchstack’s 2026 report found that 91% of new WordPress vulnerabilities in 2025 were in plugins, and a weighted median of 5 hours to first exploitation for heavily exploited ones (Patchstack). These numbers often sit next to breakage claims in blog posts. They are about attacks. They are, if anything, an argument for updating quickly, not evidence that updates break sites.
Why is there no reliable figure for WordPress update failures?
Because nobody who holds the data has published it in a form you can check, and surveys can’t produce it.
A real per-update failure rate needs two things. The first is a denominator: every update applied, across many sites. The second is a definition of “broken” that is checked the same way each time. Surveys have neither. They rely on memory, they count sites rather than updates, and the people who reply are often the ones who were burned.
Some companies do hold this kind of data. WP Engine’s Smart Plugin Manager runs updates, then checks for server errors, broken code and visual changes, and reverts the site if it finds a problem. WP Engine’s documentation says the visual regression test (a before-and-after screenshot comparison) is the primary reason Smart Plugin Manager fails an update (WP Engine). Kinsta’s Automatic Updates add-on also uses visual regression testing and rolls back failures (Kinsta). Neither page publishes how often that happens.
That WP Engine detail is worth noticing. “Failure” in an automated tool often means “the page looks different”, not “the site crashed”. A failure rate from one tool would not match a failure rate from another unless they defined failure the same way.
So how often do WordPress updates break sites?
The honest answer is: often enough that WordPress built two rollback features for it, and nobody has published a per-update rate you can check. What we can say with sources:
- In 2018, about a third of the WordPress users Pagely surveyed counted site-breaking updates among their biggest headaches.
- The WordPress core team treats a fatal error after a plugin update as the biggest update risk, and has built protections for some cases (6.3 for failed manual updates, and the automatic rollback proposed for 6.6).
- Hosts that sell automated updates test every update and roll back failures, which suggests failures are frequent enough to build products around.
Anything more precise than that, without a named source, should be treated as a guess.
How can you measure your own update failure rate?
If you manage several sites, you can produce a better number for your own sites than any survey. Keep a simple log for one quarter.
| Date | Site | Plugin or theme | From version | To version | Result | Time to fix |
|---|---|---|---|---|---|---|
| (example) | client-a.com | Example Forms | 3.1.2 | 3.2.0 | Fine | 0 |
| (example) | client-b.com | Example Builder | 5.4.0 | 5.5.0 | Visual problem on one page | 25 min |
Use four results only: fine, visual problem, something stopped working (a form, checkout, login), and fatal error. At the end of the quarter, divide problem updates by all updates for a per-update rate, and count sites with at least one problem for a per-site rate. Write down which plugins caused the problems. That list is more useful than either rate.
For the checks to run before you click Update, see how to know if a plugin update is safe. If you run many sites, TalkToWP lists pending updates across all of them and can snapshot a plugin or theme update and roll it back automatically if the site breaks. See what it checks.
Which statistic should you quote?
If you write care plan pages or client reports, quote what the source supports:
- Good: “In Pagely’s 2018 survey, 35% of WordPress users named site-breaking updates as one of their biggest headaches.”
- Good: “WordPress 6.3 added rollback for manual plugin and theme updates that fail to install, and the core team proposed automatic rollback for fatal errors in plugin auto-updates.”
- Avoid: “35% of updates break sites.” No source says this.
- Avoid: “52% of professionals had a site break in 2023.” The survey behind it can’t be found.
A client can click a link. A smaller claim with a source holds up better than a big one without. For pricing a care plan that covers this work, see WordPress maintenance pricing in 2026 and the agency maintenance checklist. If an update has already broken a site, start with there has been a critical error on this website.
Frequently asked questions
What percentage of WordPress plugin updates break sites?
There is no published, checkable per-update figure that we could find. Popular numbers come from surveys about worries or past experience, not from counting updates and failures.
Is the “52% of WordPress sites broke after an update” statistic real?
We could not confirm it. The claim appears on Usecue’s blog, attributed to “a 2023 WP survey” with no publisher or link, and we could not find the survey.
What did the Pagely 2018 WordPress survey find about updates?
In a multiple-choice question about biggest headaches, 35% of respondents picked site-breaking updates, after performance issues (52%) and security issues (41%). The page does not give the sample size.
Does WordPress roll back a plugin update that breaks the site?
Partly. Since 6.3 it restores the old version when a manual update fails to install. The automatic rollback for fatal errors in plugin auto-updates was proposed for 6.6. Neither catches broken forms or checkouts.
How can I find out how often updates break my sites?
Log every update for a quarter with its result: fine, visual problem, something stopped working, or fatal error. Divide problem updates by total updates. Note which plugins caused the problems.