DEV Community

webatla
webatla

Posted on

How Many Websites Use React? 11.05M Sites and What a Scanner Can Actually See

As of November 16, 2026, webatla counts 11,053,868 websites running React, 3.39% of the 325.8 million websites it tracks. The scan finds a version string on only about 4,500 of them.

Any answer to how many websites use React depends on what a scanner can observe in a React site, and that is worth understanding if you write or maintain React code. The live number sits on webatla's React technology page. The history tables and the full framework comparison are in the React statistics article on the webatla blog.

What a scanner can and cannot see in a React build

React renders in the browser and, in many setups, on the server as well. Either way the browser receives HTML plus JavaScript files, and React's client code takes over the page once they load. A CMS usually leaves fixed paths in the markup, and a web server often names itself in a response header. React leaves neither. Its presence has to be inferred from the JavaScript a site ships.

webatla's React page is direct about what that means. It finds React where a site exposes it through recognizable build output or development artifacts. A bundle compiled tightly enough leaves no reliable trace, and such sites stay out of the file. Two rules follow for anyone reading the number.

  • A detection is strong evidence that React is there.
  • A missing detection is weak evidence that it is not.

So 11,053,868 is a floor. The page also says the detected population leans toward sites built with default tooling, since aggressive optimization pipelines are exactly what hides the library.

There is a timing detail too, and it matters when you look at single rows. Each row carries the date its technologies were last checked. On the five preview rows of the React page those dates run from October 17 to November 11, 2026, while the file itself is dated November 16. A row describes what the scan saw on its check date. If a team shipped a rewrite last week, its row can still show the old stack.

Three numbers in a row: 11,053,868 sites with React detected, about 4,500 showing any version string, 1,503 sites with a string that follows React's numbering

Of 11,053,868 React detections, about 4,500 come with a version string, and 1,503 of those sites show a common value that follows React's own numbering.

Two checks you can run on any site

If you want to know whether one particular site runs React, you do not need a scanner. Two quick signals are enough to look for yourself.

The first is a header check, and it mostly teaches you where React is not.

curl -sIL https://example.com | grep -iE '^(server|x-powered-by):'
Enter fullscreen mode Exit fullscreen mode

The -L flag follows redirects and prints the headers of every hop, so read the last block. React sets no response header of its own, even when it renders on the server, so no header names it. You will often see a server name, and sometimes the framework wrapped around React. A Next.js server, for example, sends X-Powered-By: Next.js unless the team switched that header off. React itself never shows up there.

The second check runs in the browser. Open the site, open DevTools and paste this into the console.

// React keeps internal references on the DOM elements it manages.
const managed = [...document.querySelectorAll('*')].filter((el) =>
  Object.keys(el).some((k) => k.startsWith('__reactFiber$') || k.startsWith('__reactInternalInstance$'))
)
console.log('Elements managed by React:', managed.length)
console.log('Global React version:', window.React?.version ?? 'not exposed')
console.log('data-reactid elements (React 15 or older):', document.querySelectorAll('[data-reactid]').length)
Enter fullscreen mode Exit fullscreen mode

React 17 and later attach a property starting with __reactFiber$ to the DOM elements they render or hydrate. React 15 and 16 used __reactInternalInstance$. Older versions wrote data-reactid attributes into the markup. A non-zero count is close to proof. The version line only prints a number when React was loaded as a global script, which modern bundled apps rarely do. The React Developer Tools extension gives the same answer with less typing, because its icon lights up on pages that use React.

A zero from this snippet has the same weakness as a miss in the scan. A page that mounts React only after a click, or renders it inside an iframe, can come back empty.

Version strings: about 4,500 sites out of 11 million

The React page lists the most common version values found next to React. Only about 4,500 React sites come with any version string at all, 0.04% of the count. The version number usually sits as plain text somewhere inside the minified bundle, so the gap says more about what a scan records than about what a site hides.

The values themselves need a sanity check before anyone counts them. React's version numbers ran from 0.x up to 0.14 and then jumped to 15, followed by 16, 17, 18 and 19. There was never a React 1.x or 6.x. That rules out a large part of the list.

  • 19627.0 is the most common value of all, on 1,874 sites.
  • 33058.93, 74932.10647, 1.4.2, 0.0.2, 6.3.8, 6.5.8 and 2.14.9 follow further down.
  • 18.3.24, on 41 sites, looks like a React 18 release, but the 18.3 line only ever had 18.3.0 and 18.3.1.
  • Together these nine strings cover 2,680 sites, more than half of the rows with a version.

The public page does not say where those strings come from. The lesson applies to any scanner output: validate version fields against the real release history before you build a chart on them.

Lollipop chart of React version strings by release line: 19.x 141 sites, 18.x 956, 17.x 53, 15.x 218, 0.12 to 0.14 135, and 2,680 sites with strings outside React numbering

Within the 25 most common version values, 18.x leads with 956 sites, while 2,680 sites carry strings that React never used as release numbers.

Grouped by the major number in the string, the plausible values split like this.

Release line Sites Share of the 1,503
19.x 141 9.4%
18.x 956 63.6%
17.x 53 3.5%
15.x 218 14.5%
0.12 to 0.14 135 9.0%

No 16.x value made the top 25. For maintainers the interesting rows sit at the bottom of the table. Builds from React 15 and the 0.x line add up to 353 sites, more than twice the 141 sites with 19.x strings. Moving such a site to current React is a long trip. React.createClass and React.PropTypes left the core package in React 16, and ReactDOM.render, the classic way to mount an app, is gone in React 19. For a 0.13.3 or 15.6.1 build that looks more like a rebuild than a version bump.

Keep the scale in mind. These are a few hundred sites that happen to show a version, out of 11 million. They prove that old builds are still online, but not what share of all React sites runs old code.

Why sites leave the React count

Over the last 12 months 3.99 million sites came into the React count in webatla's history model and 2.78 million left, which works out to about 7,600 departures a day. The React page splits those departures in two.

  • Dropped it. The site is still online, but the scan no longer finds React there. 1.04 million sites.
  • Shut down. The site itself went offline. 1.74 million sites.

The history model gives the same split for every year. The chart divides each kind of exit by the React count at the start of that year.

Dumbbell chart by year from 2016 to 2025 comparing the share of React sites that removed React while staying online with the share that shut down, as a percentage of the start-of-year count

Sites removing React stayed between 6.3% and 8.5% of the start-of-year count from 2022 to 2025, while shutdowns rose from 7.8% to 15.6%.

From 2022 to 2025 the share of sites that removed React moved only between 6.3% and 8.5% a year. Shutdowns climbed from 7.8% to 15.6% over the same years. In the first ten months of 2026 shutdowns already reached 14.6% of the count at the start of the year, and removals 9.1%. Over the last 12 months 1.04 million sites removed React out of 9.84 million at the end of October 2025, about 10.6%, which is above every year from 2022 to 2025. The high removal rates of 2018 and 2021 come with single months of very large drops (June and September 2018, July 2021), which may be a measurement effect rather than a wave of rewrites. Early years of the model are approximate.

So the growing number of exits is mostly sites going offline. When a React site drops out of the count, the usual story is a project that ended. A team pulling React out of a live product is the smaller group. Even that group is not all real migrations. A "dropped it" row can also be a site whose new build no longer exposes React to the scan, and the public pages cannot tell the two apart.

Are existing sites moving away from React?

Switching is the better measure of developer preference. A switch is an existing site that adds or removes a technology while it stays online. Over the last 12 months 1.16 million existing sites added React and 1.04 million removed it. Summed from the monthly rows, the balance is about 126,000, or 1.3 for every 100 sites React had at the end of October 2025.

The same arithmetic works for any framework. The chart applies it to React's three direct rivals (Vue.js, Angular and Svelte) and four older libraries (AngularJS, Backbone.js, Prototype and jQuery). Tools built on React, such as Next.js, are left out, since their own webatla pages tie them to React.

Diverging lollipop chart of switching balance per 100 sites over the last 12 months: Svelte plus 21.4, Angular plus 9.2, React plus 1.3, Vue.js plus 0.3, jQuery minus 3.3, Prototype minus 6.9, AngularJS minus 10.5, Backbone.js minus 11.5

Switched to it minus dropped it over the last 12 months, per 100 sites at the end of October 2025. All four older libraries lost existing sites.

The older libraries all lose existing sites on balance: Backbone.js 11.5 per 100, AngularJS 10.5, Prototype 6.9 and jQuery 3.3. React gains a little. Vue.js sits close to zero, with 529,000 sites switching to it and 524,000 dropping it.

Svelte and Angular post the largest positive rates, but on small bases. Svelte's +21.4 per 100 is about 11,900 sites, roughly a tenth of React's balance in absolute terms. Angular's +9.2 is about 13,000. Today's files count Vue.js on 2,061,594 sites, Angular on 156,514 and Svelte on 84,502. React alone holds 46.3% of the detections in webatla's JavaScript frameworks category (React market share across 54 technologies, where a site with two frameworks counts once for each), so its modest rate still adds more existing sites than any other framework in the chart.

Both directions are large for every framework in the chart. Some of that back and forth may be measurement, such as a site whose build changed between two checks. The balance is the steadier number to quote.

Existing React sites are not leaving on net. The libraries losing existing sites are the ones that came before React.

Domain extension or server country: two different questions

Every row in the React file has two location fields, and they answer different questions. Country is where the server is, found by IP address. The domain extension is where the owner chose to register. For React sites the two disagree a lot.

89.1% of React sites with a known server country sit on US servers, 9,671,733 of them. Put each country's domain extension next to the React sites hosted inside that country and the gap becomes visible.

Dumbbell chart comparing React sites hosted in a country with React sites on that country's domain extension, for the UK, Australia, Brazil, Canada, France, India, Russia, Germany and the Netherlands

React sites under .uk outnumber React sites on UK servers 431,186 to 18,579. Germany and the Netherlands host more React sites than their own extensions hold.

The .uk extension holds 431,186 React sites, while UK servers host 18,579, about 23 to 1. Australia (.au, 167,920 against 11,876) and Brazil (.br, 237,447 against 19,040) come out at about 14 to 1 and 12 to 1. Germany and the Netherlands run the other way. German servers host 417,973 React sites against 269,953 under .de, and Dutch servers 193,845 against 75,980 under .nl. So German and Dutch servers carry many React sites under .com and other extensions.

The most likely reason for the pull toward US servers is platform hosting. A site builder or front-end cloud serves its customers from its own infrastructure, wherever the customer lives. Wix fits that pattern, with React in all five of its sample rows and 99.4% of its sites on US servers. A sizable share of the 11 million React sites are therefore likely builder pages whose owners never wrote a line of React.

For your own queries this means picking the field that matches the question. Looking for React sites aimed at British users? Filter by extension. Counting React sites served from EU servers, for latency or data residency? Filter by country. The two filters return very different lists.

What the numbers mean if you build or maintain React apps

  • The count is bigger than the number of React codebases. One team can run React on many domains, and builder pages likely add a share of rows that need no React developer. The public pages give no exact overlap, so how many separate React projects sit behind the 11 million stays unmeasured.
  • The base keeps growing, with heavy turnover. React usage statistics on the technology page show +35% from October 2023 to October 2026 and +12% over the last 12 months. In October 2026 alone the history model has 285,000 sites coming in and 171,000 leaving, so any saved list of React sites ages quickly.
  • Exits are mostly shutdowns. Removal rates stayed in a narrow band from 2022 to 2025 and rose in 2026, but switch-ins still outweigh removals. Nothing in the data looks like a rush away from React.
  • Old builds are still online. The visible version strings include hundreds of sites on 15.x and 0.x code. If you inherit one of them, plan for a rebuild.
  • Single-site answers need a direct check. A row reflects one check date and one view of the shipped code. When it matters, run the console snippet above.

Quick answers for developers

Do 11 million React sites mean 11 million React projects?

No. The count is one row per domain where React was detected. Site builders that ship React for their customers likely add many of those rows, and one team can run React on many domains.

Can I tell which React version a site runs?

Sometimes. window.React.version works only when React loads as a global script. With the React Developer Tools extension installed, the renderer reports its version to the extension, so the version is often readable on a site you check by hand. In webatla's data about 4,500 React sites expose any version string, and many of those strings are not React release numbers.

Are sites leaving React for Vue.js or Svelte?

Not on net. Over the last 12 months 1.16 million existing sites added React and 1.04 million removed it. Svelte and Angular gained existing sites faster in relative terms, but from bases a fraction of React's size.

How fresh is the count?

The technology page is rebuilt every day. Individual rows carry their own last-checked date, so a single site's entry can be a few weeks older than the file.

Sources

Top comments (0)