Analysis · 12 November 2026

WordPress has dropped to 39.2% of websites in W3Techs’ data, and most of the fall happened this year

For three years W3Techs’ count of websites running WordPress barely moved: 43.1% in October 2023, a peak of 43.6% on 1 January 2025, and still 43.0% on 1 January 2026. On 12 November 2026 it reads 39.2%. That is 4.4 percentage points off the peak, and 3.8 of them went in the last ten and a half months, with each month’s fall larger than the one before. The figure is a share of the websites W3Techs surveys, not a count of WordPress installations, and the data itself has an odd feature: the share WordPress lost this year did not go to Shopify, Wix or any other system W3Techs tracks. It went to the row labelled “None”. This article sets out the numbers, our infographic of them, what the survey can and cannot tell you, and what the trend changes for someone who runs or is leaving WordPress, which for security is less than the headline suggests.

Latest reading
39.2% of all websites W3Techs surveys run WordPress, as of 12 November 2026
Peak
43.6% on 1 January 2025; the figure has fallen at every quarterly snapshot since
Fall from peak
4.4 percentage points, about a tenth of the peak share; 3.8 of those points since 1 January 2026
Where it went
Sites with none of the monitored CMSs rose from 28.6% to 32.3% in 2026; the other tracked systems together were roughly flat
Still the largest
WordPress is on 57.9% of the websites whose CMS W3Techs can identify; Shopify, second, is on 5.4% of all sites
What is measured
A share of a survey sample of websites, detected from the outside. Not installations, downloads or active sites

Start here

Three flat years, then ten months of accelerating decline

W3Techs publishes a daily survey of which technologies websites use, and its content management figures are the ones most often quoted when someone says WordPress runs some fraction of the web. Its quarterly history shows WordPress between 43.0% and 43.6% of all surveyed websites at every snapshot from October 2023 to January 2026. From there the line bends: 42.5% on 1 April 2026, 41.5% on 1 July, 40.2% on 1 October and 39.2% on 12 November.

Three things about that fall are worth holding on to before reading any explanation of it. It is recent: nine tenths of the drop from the peak happened in 2026. It is accelerating: the monthly fall grew from 0.1 or 0.2 points early in the year to 0.7 points during October. And it is unusually one-sided: over the same months the tracked competitors gained almost nothing between them, while the share of sites on which W3Techs detects no monitored CMS rose by nearly the same amount WordPress lost.

None of that makes WordPress small. It is still on more than seven times as many surveyed sites as the next system. What it does mean is that a number the WordPress world has treated as a fixed point for years is moving, quickly, and nobody has yet published a tested explanation of why.

Figure 1 · WordPress usage at a glance

Flat for three years, then 3.8 points in ten months — and the share went to “None”.

Share of all websites1 Oct 20231 Jan 20251 Jan 202612 Nov 2026Change in 2026
WordPress43.1%43.6% (peak)43.0%39.2%−3.8 points
None of the monitored CMSs31.6%29.6%28.6%32.3%+3.7 points
Shopify4.0%4.7%5.0%5.4%+0.4 points
Wix2.5%3.3%4.2%4.2%No change
Joomla and Drupal together2.9%2.5%2.1%1.7%−0.4 points
Share of all websites in W3Techs’ survey at four dates, and the change during 2026. WordPress lost 3.8 points this year; the “None” row gained 3.7; Shopify and Wix together gained 0.4.

The infographic

WordPress has given back four points since its 2025 peak

Our chart of W3Techs’ quarterly history, October 2023 to 12 November 2026. Every value drawn on it appears in the tables further down this page.

In the infographic below, the upper chart is the WordPress line alone. The lower chart compares where each system stood on 1 October 2023 with where it stands now, in percentage points of all surveyed websites, so the bars can be added and compared directly. Over those three years WordPress is down 3.9 points, Wix is up 1.7 and Shopify is up 1.4. Read on its own, that looks like a straightforward story of hosted builders taking share from WordPress. The month-by-month figures below show why it is not that simple.

Infographic titled “WordPress has given back four points since its 2025 peak”. A line chart of the share of all websites running WordPress according to W3Techs shows it near 43% from October 2023, peaking at 43.6% on 1 January 2025, then falling to 43.0% in January 2026, 42.5% in April, 41.5% in July, 40.2% in October and 39.2% on 12 November 2026, a change of minus 4.4 points from the peak. A bar chart below shows the change in share from October 2023 to November 2026 for thirteen categories: WordPress minus 3.9 points, Joomla minus 0.7, Drupal minus 0.5, Adobe Systems minus 0.5, Google Systems minus 0.2, Webflow plus 0.1, GoDaddy Website Builder plus 0.2, Squarespace plus 0.3, Duda plus 0.3, Tilda plus 0.4, No CMS plus 0.7, Shopify plus 1.4 and Wix plus 1.7.
Infographic · WordPress usage, October 2023 to November 2026

Share of all websites running WordPress, and the change in share for each system W3Techs lists, from its quarterly history. The vertical axis starts at 38% so the change is visible; drawn from zero, the line would fall from 43.6 to 39.2 on a scale of 50. Source: W3Techs, data as of 12 November 2026.

The quarterly record

Every snapshot since October 2023

Percentages of all websites in W3Techs’ survey, taken on the first day of each quarter, with the latest reading on 12 November 2026. “None” is W3Techs’ row for websites that use none of the content management systems it monitors.

Two periods are visible in the table below. Through 2024 and 2025 the “None” column falls steadily, from 31.6% to 28.6%, while WordPress holds and Wix and Shopify grow: sites that had no detected CMS were, on balance, being replaced in the sample by sites that had one. In 2026 the direction reverses. “None” climbs back past where it started, and WordPress falls by almost exactly the amount “None” rises.

Because 100% minus “None” is the share of sites with an identified CMS, the table also gives WordPress’s share among those sites. W3Techs states it as 57.9% today. Applying the same arithmetic to the 1 January 2025 row gives about 61.9%, so WordPress has lost ground on that measure too, but more slowly: four points of a base in the sixties instead of four points of a base in the forties.

SnapshotWordPressNoneShopifyWixSquarespace
1 Oct 202343.1%31.6%4.0%2.5%2.1%
1 Jan 202443.1%31.4%4.1%2.6%2.1%
1 Apr 202443.3%31.0%4.3%2.6%2.1%
1 Jul 202443.4%30.8%4.4%2.8%2.1%
1 Oct 202443.5%30.4%4.5%3.0%2.2%
1 Jan 2025 (peak)43.6%29.6%4.7%3.3%2.2%
1 Apr 202543.5%29.1%4.8%3.6%2.3%
1 Jul 202543.4%28.8%4.8%3.9%2.4%
1 Oct 202543.3%28.6%4.8%4.1%2.4%
1 Jan 202643.0%28.6%5.0%4.2%2.4%
1 Apr 202642.5%28.9%5.1%4.3%2.5%
1 Jul 202641.5%29.9%5.2%4.3%2.5%
1 Oct 202640.2%31.5%5.4%4.2%2.4%
12 Nov 202639.2%32.3%5.4%4.2%2.4%

2026, month by month

Each month’s fall has been larger than the last

W3Techs’ monthly history, first of each month, for the two rows that moved. The change columns are the difference from the previous row, in percentage points.

From 1 January to 12 November 2026 WordPress lost 3.8 points and “None” gained 3.7. The monthly steps track each other closely and both grow through the year: WordPress fell 0.5 points over the whole first quarter and 0.7 points in October alone. The eleven days to 12 November took another 0.3 from WordPress, which is a pace similar to October’s; “None” moved only 0.1 in those days, so the two do not mirror each other at every step.

Whether this counts as sudden depends on the scale you read it at. There is no single-day cliff in the data, which is what a one-off reclassification would usually produce. There is a curve that was flat for three years and has steepened every quarter since January. On a figure that had moved within a 0.6-point band since late 2023, that is a sharp change.

SnapshotWordPressChangeNoneChange
1 Jan 202643.0%28.6%
1 Feb 202642.8%−0.228.6%0.0
1 Mar 202642.7%−0.128.8%+0.2
1 Apr 202642.5%−0.228.9%+0.1
1 May 202642.2%−0.329.2%+0.3
1 Jun 202641.9%−0.329.5%+0.3
1 Jul 202641.5%−0.429.9%+0.4
1 Aug 202641.2%−0.330.4%+0.5
1 Sep 202640.7%−0.530.9%+0.5
1 Oct 202640.2%−0.531.5%+0.6
1 Nov 202639.5%−0.732.2%+0.7
12 Nov 2026 (11 days)39.2%−0.332.3%+0.1

Where the share went

Not to Shopify or Wix: to the “None” row

If site owners were moving from WordPress to competing platforms at this rate, the competitors’ lines would show it. They do not. Between 1 January and 12 November 2026, the eleven other systems W3Techs lists moved as follows: Shopify gained 0.4 points; Wix, Squarespace, Tilda, Duda and GoDaddy Website Builder did not change; Joomla and Drupal each lost 0.2; Webflow, Adobe’s systems and Google’s systems each lost 0.1. Added together that is a net loss of about 0.3 points across everything that is not WordPress.

Several of the hosted builders have in fact slipped slightly since midsummer. Wix was 4.3% on 1 July and is 4.2% now. Squarespace went from 2.5% to 2.4%, Webflow from 0.9% to 0.8%. These are one-step moves at the edge of rounding, but they point the same way: in the second half of 2026 the share of surveyed sites with any identified CMS is shrinking, and WordPress, as by far the largest, accounts for nearly all of the shrinkage in absolute terms.

The three-year comparison in the infographic remains true. Wix and Shopify are up 1.7 and 1.4 points since October 2023. But almost all of that growth happened in 2024 and 2025, while WordPress was flat. The period in which WordPress fell is not the period in which they grew.

System1 Jan 202612 Nov 2026Change in 2026
WordPress43.0%39.2%−3.8
None of the monitored systems28.6%32.3%+3.7
Shopify5.0%5.4%+0.4
Wix4.2%4.2%0.0
Squarespace2.4%2.4%0.0
Joomla1.3%1.1%−0.2
Drupal0.8%0.6%−0.2
Webflow0.9%0.8%−0.1
Tilda0.8%0.8%0.0
Duda0.7%0.7%0.0
GoDaddy Website Builder0.6%0.6%0.0
Adobe Systems0.7%0.6%−0.1
Google Systems0.6%0.5%−0.1

Reading it correctly

What W3Techs measures, and three explanations nobody has tested

The headline number is easy to misread as “a tenth of WordPress sites have disappeared”. The survey does not say that, and cannot.

W3Techs describes its sample as “well over 20 million sites”, limited to what it calls the relevant web: sites with meaningful content or functionality, ranked using Google’s Chrome User Experience Report and a customised Tranco list. It counts websites, not pages or subdomains, and it leaves out redirected domains, duplicates and sites showing only a default server page. A CMS is identified from the outside, by matching patterns in the HTML, scripts, headers, cookies and URL structure, and W3Techs says plainly that no survey of this kind can be fully accurate and that a site can choose to hide most of its technologies.

So the figure is a share of a sample, read from outside. It can fall because WordPress sites closed or moved. It can equally fall because the sample gained sites that were never WordPress, because the mix of which sites qualify as relevant changed, or because sites that still run WordPress stopped looking like it from outside. A share also says nothing about absolute numbers: if the surveyed web is growing, WordPress’s share can fall while the number of WordPress sites holds steady.

Explanation 1: more sites built without a CMS

If a growing proportion of new sites are static pages, custom applications or output from site generators W3Techs does not classify as a CMS, they enter the sample under “None” and dilute every tracked system in proportion to its size. This fits the pattern of WordPress falling most while smaller systems slip slightly. We have no measurement that confirms it.

Explanation 2: the sample changed

W3Techs draws its relevant web from external ranking data. A change in which sites those rankings include would change the denominator without any site changing its software. W3Techs’ methodology page records one such change, the move away from Alexa rankings in November 2022, and nothing since. The gradual, accelerating shape argues against a single switch-over.

Explanation 3: WordPress that no longer looks like WordPress

Headless builds, aggressive caching and CDN layers, and hardening that strips version strings and generator tags all remove the markers a scanner looks for. Sites detected as “None” for this reason are still WordPress installations with the same exposure. How many there are is unknown.

What would settle it

A statement from W3Techs on what makes up the growing “None” group, or agreement from sources that count differently: WordPress.org’s own download and active-version statistics, and crawl-based datasets such as the HTTP Archive. We have not compared those here, and this article does not claim to know the cause.

What it means for security

A smaller share does not make a smaller target

From where we sit, the practical question is whether any of this changes the risk to a site that runs WordPress. It does not, in either direction. Attackers do not choose targets from a market-share chart. They scan for a specific vulnerable plugin, theme or Core version across every address that answers, and a plugin with a million active installations is exactly as attractive at 39% as it was at 43%. Nothing in the weekly briefings on this site, the Core update-window flaw, the exploited plugin list, the PHP and kernel chains underneath, depends on WordPress’s share of the web.

The same W3Techs page carries a figure that matters more to an operator than the headline. Of the sites it identifies as WordPress, 69.0% are on version 7, 24.3% on version 6, 4.6% on version 5, 1.9% on version 4 and 0.2% on version 3. Nearly a third of detected WordPress sites are a major version or more behind, and 6.7% are on version 5 or older. Whatever is happening to the total, that is the population in which most compromises start.

There is one way a declining share could raise risk over time, and it is worth watching without overstating: ecosystems that shrink tend to accumulate abandoned software. Plugins whose authors move on stop receiving fixes long before their installations disappear. That is already the most common reason we find an unpatched flaw on a site, and it needs no market trend to be a problem today. The defence is the same as ever: know what is installed, remove what is unmaintained, and do not rely on an update that will never come.

If you are staying on WordPress

Nothing here is a reason to move. Keep Core, plugins, themes and the PHP beneath them current, remove plugins that have had no release in a year or more, and check that every administrator account is one you created. The share figure changes none of those tasks.

If you run sites for clients

Expect the number to come up in conversations. The accurate answer is that one survey shows a four-point fall in share with no established cause, that WordPress remains the largest platform by a wide margin, and that the security of a given site depends on how it is maintained, not on how many others use the same software.

If you are hiding the fingerprint

Removing the generator tag and version strings keeps a site out of some surveys and some lazy scans. It does not stop a scanner that requests a known vulnerable plugin path directly, which is how most mass exploitation works. Treat it as tidiness, not protection.

If you are moving away

The old installation is the risk, not the new platform. A WordPress site that has been replaced but not removed keeps running, stops being updated, and stays reachable. The next section is the list for that.

If you are leaving WordPress

Decommission the old site properly, or it becomes the way in

Some of the sites behind this trend have genuinely moved. A migration that ends with the old WordPress still on the server, on a subdomain, or in a backup directory leaves an unmaintained application with a database of user records attached to the same hosting account.

  1. 01

    Find every copy, not just the live one

    List the installations on the account before anything is switched off: the production site, staging and development copies, an old “/blog” or “/shop” directory, subdomains created for a redesign, and copies left by a previous agency. Each one has its own files, database and administrator accounts, and each is reachable unless someone has made it otherwise.

  2. 02

    Export what you must keep, then check it

    Take the content export, the media library and a final database dump, and confirm they open and are complete before removing anything. Store the dump somewhere that is not the web root: a database export left in a public directory is a download link to every user’s email address and password hash.

  3. 03

    Remove the application, not just the link to it

    Pointing the domain at the new platform does not delete the old site if it stays reachable at the server’s address or on another hostname. Delete the WordPress files and drop the database and its database user. If the installation has to be kept for reference, put it behind authentication at the web-server level and stop it executing PHP for the public.

  4. 04

    Close the accounts and credentials around it

    Delete the SFTP, SSH and hosting-panel users that existed for the old site, rotate any API keys and SMTP credentials stored in its configuration, and cancel premium plugin licences so the keys stop being valid. Remove scheduled jobs that called the old site.

  5. 05

    Tidy DNS and redirects

    Remove DNS records for hostnames that no longer serve anything, so they cannot be claimed by someone else on shared infrastructure. Redirect the old URLs you care about from the new platform, and check a few weeks later that nothing still resolves to the old server.

  6. 06

    If it shared a server, check the neighbours

    On shared or reseller hosting, an abandoned WordPress site is a common starting point for an infection that then spreads to the other sites under the same account. If the old site sat unpatched for months before it was removed, review the sites that lived beside it.

Sources and method

Where every number in this article comes from

All usage figures are W3Techs’, read on 12 November 2026 from its quarterly and monthly histories of content management system usage, its WordPress detail page and its content management overview. They are quoted to the one decimal place W3Techs publishes and were not independently re-measured. The quarterly values are W3Techs’ snapshots for the first day of each quarter; the final column on both history pages is dated 12 November 2026.

The changes, sums and the roughly-a-tenth figure are our arithmetic on those published values and inherit their rounding. The 61.9% share among identified-CMS sites for January 2025 is calculated from the WordPress and “None” rows using the relationship that reproduces W3Techs’ own 57.9% for today; W3Techs’ market-share history was not read. The description of the survey’s sample, counting rules and detection method is from W3Techs’ technologies and FAQ pages, quoted or closely paraphrased.

The infographic was drawn by us from the quarterly table. Its line chart uses a vertical axis starting at 38%, which makes a four-point change look steep; the caption says so. The three explanations in this article are possibilities consistent with the data and are labelled as untested. This is one survey. WordPress.org’s statistics, the HTTP Archive and other crawlers measure different populations in different ways and were not compared here.

Common questions

Common questions about the WordPress usage figures

Has the number of WordPress installations fallen by a tenth?

The data does not say that. W3Techs reports the share of websites in its survey sample on which it detects WordPress, and that share has fallen from 43.6% to 39.2%. A share can fall because sites left WordPress, because the sample gained sites that never used it, or because WordPress sites became harder to detect. W3Techs does not publish a count of installations, and this article does not estimate one.

What percentage of websites use WordPress now?

According to W3Techs on 12 November 2026, 39.2% of all websites in its survey, and 57.9% of the websites whose content management system it can identify. The peak was 43.6% of all websites on 1 January 2025.

When did the decline start?

The quarterly figure has fallen at every snapshot since the 1 January 2025 peak, but by only 0.1 points a quarter through 2025. The steep part began in 2026: the figure was 43.0% on 1 January 2026 and has dropped 3.8 points since, with the monthly fall growing from 0.1–0.2 points early in the year to 0.7 points in October.

Are people moving from WordPress to Shopify or Wix?

Not on the evidence of this data for 2026. Shopify gained 0.4 points this year and Wix none, against a WordPress loss of 3.8. Both grew substantially in 2024 and 2025, when WordPress was flat. The category that grew in 2026 is “None”: websites using none of the systems W3Techs monitors, up 3.7 points.

Why is it falling?

Nobody has established that. The published numbers are consistent with more new sites being built without a recognised CMS, with a change in the composition of the survey sample, and with WordPress sites that no longer expose the markers a scanner looks for. W3Techs has not published an explanation on the pages we read, and we have not tested any of the three against other data.

Does this make WordPress less of a target for attackers?

No. Attacks on WordPress are automated scans for specific vulnerable plugins, themes and Core versions, and their economics depend on the number of vulnerable installations, which a four-point change in survey share does not meaningfully alter. WordPress is still on more than seven times as many surveyed sites as any other system.

Should I move my site off WordPress because of this?

Not because of this. A usage survey says nothing about whether a platform suits your site. If you are moving for other reasons, the security work is in removing the old installation completely: files, database, accounts, credentials and DNS records. A replaced but still-running WordPress site is a common source of later compromise.

Is the chart exaggerating the drop?

The line chart’s axis starts at 38% so that a change of a few points can be seen at all, and the infographic says so on the chart and in its caption. On an axis from zero the same data is a line that dips from 43.6 to 39.2. Both are accurate; the tables on this page give the values without any scaling.

Choose the response path

An old WordPress site is still online: choose the response path.

Staying on WordPress

Know exactly what is installed, what is unmaintained, and what is exposed.

Best when you run one or more WordPress sites, want plugins, themes, users, configuration and the server beneath them reviewed in one pass, and want a written result with each finding ranked and an acceptance test for its fix.

Leaving, or already left

An old WordPress site still on the server, or other sites sharing an account with one.

Best when a site has been migrated but the previous installation, its staging copies or its database are still in place, or when something on the same hosting account has started behaving oddly since.