Website Statistics 2026
By Axis Intelligence Research
Co-author: James Porter | Last updated: October 18, 2026 | License: CC BY 4.0
The web counts 1,494,915,628 websites, but Axis Intelligence Research estimates only 216.7 million of them are active, 14.5%, or roughly one site in seven. Since October 2024 the dormant web grew 36.4% while the active web grew 11.8%: about 15 parked or empty hostnames for every new live site.
Quick Answer
There are 1.49 billion websites in July 2026 according to the Netcraft July 2026 Web Server Survey, spread across 305.3 million domains and 14.8 million web-facing computers. Axis Intelligence Research estimates 216.7 million are active (14.5%). Registered domain names reached 401.6 million at the end of June 2026 (Verisign DNIB Q2 2026). Only 48% of mobile origins pass Core Web Vitals, 95.9% of the top million home pages carry detectable WCAG failures, and humans generated just 47% of HTML requests on Cloudflare’s network on December 2, 2025. Our composite Web Vital Signs Index (WVSI) reads 42.2 out of 100.
Key Findings
- Axis Intelligence Research estimates that 216.7 million of the web’s 1.49 billion sites were active in July 2026, an active share of 14.5%.
- Axis Intelligence Research found that between October 2024 and July 2026 the number of non-active websites grew 36.4%, three times faster than active websites (+11.8%).
- Axis Intelligence Research calculates that 14.9 dormant hostnames were added for every new active website over the same 21 months.
- The Axis Intelligence Research Web Vital Signs Index (WVSI) records a baseline reading of 42.2 out of 100 in October 2026, held down by accessibility (4.1) and site activity (14.5).
- Axis Intelligence Research calculates that Anthropic’s lowest 2025 crawl-to-refer ratio (25,000:1) was 833 times Google’s highest (30:1), per Cloudflare Radar data.
How Many Websites Are There in 2026?
1,494,915,628. That is the hostname count in Netcraft’s July 2026 survey, published July 27, 2026, and the latest edition available as of this update. The survey added 5.5 million sites in a month. Netcraft counts every hostname that answers an HTTP request, so a single company can contribute hundreds of entries.
| Metric | July 2026 | January 2026 | October 2024 | Source |
|---|---|---|---|---|
| Websites (hostnames) | 1,494,915,628 | 1,376,952,390 | 1,131,068,688 | Netcraft |
| Domains responding | 305,348,459 | 295,088,466 | 271,754,817 | Netcraft |
| Web-facing computers | 14,772,048 | 14,365,221 | 13,003,235 | Netcraft |
| Active websites (Axis estimate) | 216.7M | 207.5M | 193.8M | Axis Intelligence Research, from Netcraft |
Sources: Netcraft July 2026, Netcraft January 2026, Netcraft October 2024.
Month by month in 2026
| Survey | Websites | Domains | Active (Axis est.) | Active share | Source |
|---|---|---|---|---|---|
| Jan 2026 | 1,376,952,390 | 295,088,466 | 207.5M | 15.07% | Netcraft |
| Feb 2026 | 1,419,977,535 | 297,071,903 | 208.2M | 14.67% | Netcraft |
| Mar 2026 | 1,427,812,919 | 297,642,950 | 208.1M | 14.58% | Netcraft |
| Apr 2026 | 1,433,742,238 | 299,194,059 | 210.5M | 14.69% | Netcraft |
| May 2026 | 1,468,295,818 | 302,162,421 | 215.6M | 14.68% | Netcraft |
| Jun 2026 | 1,489,396,284 | 304,146,307 | 217.6M | 14.61% | Netcraft |
| Jul 2026 | 1,494,915,628 | 305,348,459 | 216.7M | 14.50% | Netcraft |
From January to July 2026 the hostname count rose 8.6%. Domains rose 3.5%. Web-facing computers rose 2.8%. Active sites rose 4.4%. The web is getting wider faster than it is getting busier.
How many hostnames per domain?
Axis Intelligence Research calculates 4.9 hostnames per responding domain (1,494,915,628 ÷ 305,348,459) and 6.9 hostnames per active site (1,494,915,628 ÷ 216,702,417). Subdomain sprawl from SaaS tenants, preview deployments and wildcard DNS inflates the headline count; the domain figure is the steadier unit for trend work.
James Porter: Read 1.49 billion as “things answering on port 80 or 443”, not “businesses online”. Every wildcard DNS record on a shared host is a site to a crawler. If you are sizing a market, start from domains or active sites. Never from hostnames.
How Many Active Websites Are There?
About 216.7 million, or 14.5% of all sites (Axis Intelligence Research estimate, July 2026). Netcraft does not publish the size of its active-sites universe directly. It publishes each vendor’s active-site count and share, which lets the total be recovered:
Active sites = Cloudflare active sites ÷ Cloudflare active share = 50,838,387 ÷ 0.2346 = 216,702,417
Netcraft defines active sites by excluding hostnames whose content is auto-generated or duplicated, such as parked domains and placeholder pages. The same formula on the January 2026 survey gives 207.5 million (41,709,809 ÷ 0.2010) and on October 2024 gives 193.8 million (31,263,058 ÷ 0.1613). This value matches the active-web figure on our web hosting statistics page; both pages draw from the same registry entry.
The dormant web is outgrowing the live web
| Period: Oct 2024 → Jul 2026 | Oct 2024 | Jul 2026 | Change | Source |
|---|---|---|---|---|
| All websites | 1,131.1M | 1,494.9M | +32.2% | Netcraft |
| Active websites (Axis est.) | 193.8M | 216.7M | +11.8% | Axis Intelligence Research |
| Non-active websites (Axis est.) | 937.2M | 1,278.2M | +36.4% | Axis Intelligence Research |
Active sites added: 22.9 million. Non-active hostnames added: 341.0 million. Ratio: 14.9 dormant hostnames per new active site. The active share slid from 17.14% to 14.50% in 21 months.
Who serves the active web?
| Web server | Active sites (Jul 2026) | Share | Source |
|---|---|---|---|
| Cloudflare | 50,838,387 | 23.46% | Netcraft |
| nginx | 31,304,892 | 14.45% | Netcraft |
| Apache | 26,229,283 | 12.10% | Netcraft |
| 22,443,103 | 10.36% | Netcraft |
Across all 1.49 billion sites the order differs: nginx 20.37%, Cloudflare 16.67%, Apache 11.50%, OpenResty 8.02%. In Netcraft’s top million sites Cloudflare leads at 28.29%, ahead of nginx (20.02%) and Apache (15.56%). Cloudflare’s active share climbed from 16.13% in October 2024 to 23.46% in July 2026. Counted by machines rather than sites, nginx runs 43.92% of web-facing computers.
James Porter: The busier the site, the more likely a proxy sits in front of it. Cloudflare’s lead grows as you move from all sites to active sites to the top million. That is a reverse-proxy story, not a web-server story. The origin behind the edge is often still nginx or Apache, and the survey cannot see it.
How Many Domain Names Are Registered?
401.6 million across all top-level domains at the end of Q2 2026 (June 30, 2026), per the Verisign Domain Name Industry Brief. That is 9.1 million more than Q1 (+2.3%) and 29.9 million more than a year earlier (+8.1%).
| Metric | Value | As of | Source |
|---|---|---|---|
| Domain registrations, all TLDs | 401.6M | 2026-06-30 | Verisign |
| Quarter-over-quarter change | +9.1M (+2.3%) | Q2 2026 | Verisign |
| Year-over-year change | +29.9M (+8.1%) | Q2 2026 | Verisign |
| Domains answering HTTP (Netcraft) | 305.3M | 2026-07 | Netcraft |
| Responding share of registrations (Axis calc.) | ≈76% | 2026 | Axis Intelligence Research |
Axis Intelligence Research divides Netcraft’s responding domains by Verisign’s registrations (305,348,459 ÷ 401,600,000 = 76.0%). The two counts use different methods, so treat the result as an order of magnitude: roughly a quarter of registered names serve nothing a crawler can see.
What Share of Websites Pass Core Web Vitals?
48% of mobile origins and 56% of desktop origins had good Core Web Vitals in July 2025, per the Web Almanac 2025 performance chapter, built on Chrome UX Report field data. Mobile was at 32% in 2021.
| Year | Mobile origins, good CWV | Desktop origins, good CWV | Source |
|---|---|---|---|
| 2021 | 32% | 41% | HTTP Archive |
| 2022 | 31% | 44% | HTTP Archive |
| 2023 | 36% | 48% | HTTP Archive |
| 2024 | 44% | 55% | HTTP Archive |
| 2025 | 48% | 56% | HTTP Archive |
Which metric fails most?
| Metric (July 2025) | Mobile good | Desktop good | Source |
|---|---|---|---|
| Largest Contentful Paint (LCP) | 62% | 74% | HTTP Archive |
| Interaction to Next Paint (INP) | 77% | 97% | HTTP Archive |
| Cumulative Layout Shift (CLS) | 81% | 72% | HTTP Archive |
| Time to First Byte (TTFB, diagnostic) | 44% | — | HTTP Archive |
LCP is the binding constraint on mobile. INP improved from 74% to 77% in a year. Median mobile Total Blocking Time in lab tests reached 1,916 ms.
Do popular sites perform better?
Not in the middle of the ranking. Mobile pass rates are 51% for the top 1,000 origins, 42% for the top 10,000, 37% for the top 100,000, 42% for the top million and 48% for the top 10 million. Home pages trail inner pages: 45% versus 56% on mobile. For platform-level results, see our WordPress statistics.
James Porter: The top-100k dip is the ad-tech belt. Big enough to carry three tag managers, too small to fund a performance team. Check your TTFB before your JavaScript. If the first byte is slow, nothing downstream saves LCP.
How Big Is the Average Web Page?
2,559 KB for the median mobile home page and 2,862 KB on desktop (July 2025), per the Web Almanac 2025 page weight chapter. The mobile home page weighed 845 KB in July 2015: a 202.8% increase in a decade, and +8.4% in the last year alone.
| Resource (median mobile home page) | Bytes | Source |
|---|---|---|
| Images | 911 KB | HTTP Archive |
| JavaScript | 632 KB | HTTP Archive |
| Fonts | 122 KB | HTTP Archive |
| CSS | 77 KB | HTTP Archive |
| HTML | 22 KB | HTTP Archive |
Inner pages are lighter: 1,769 KB on mobile, 1,963 KB on desktop. The median page makes 72 requests on mobile and 77 on desktop. 98.1% of pages request JavaScript.
Verification note: the chapter’s narrative text gives 2,362 KB for the 2025 mobile home page, while its figures give 2,559 KB. Only 2,559 KB reproduces the stated 202.8% decade growth from 845 KB, so the dataset uses 2,559 KB.
James Porter: 22 KB of HTML carrying 2.5 MB of everything else. Ship the images in AVIF or WebP, set explicit dimensions, and stop lazy-loading the hero. Those three changes fix more pages than any framework migration.
Are Websites Accessible?
No, overwhelmingly. The WebAIM Million 2026 found detectable WCAG 2 failures on 95.9% of the top one million home pages in February 2026, up from 94.8% in 2025. Only 4.1% showed no automatically detectable failures, and automated tools catch only part of the problems, so true conformance is lower.
| Year | Home pages with detected WCAG failures | Source |
|---|---|---|
| 2019 | 97.8% | WebAIM |
| 2020 | 98.1% | WebAIM |
| 2021 | 97.4% | WebAIM |
| 2022 | 96.8% | WebAIM |
| 2023 | 96.3% | WebAIM |
| 2024 | 95.9% | WebAIM |
| 2025 | 94.8% | WebAIM |
| 2026 | 95.9% | WebAIM |
The six errors that dominate
| Error type (Feb 2026) | Share of home pages | Source |
|---|---|---|
| Low-contrast text | 83.9% | WebAIM |
| Missing image alt text | 53.1% | WebAIM |
| Missing form input labels | 51.0% | WebAIM |
| Empty links | 46.3% | WebAIM |
| Empty buttons | 30.6% | WebAIM |
| Missing document language | 13.5% | WebAIM |
WebAIM detected 56,114,377 errors, an average of 56.1 per page (51 in 2025), while pages grew to 1,437 elements (+14.3%). Pages using ARIA averaged 59.1 errors against 42 for pages without it.
By sector and platform
| Segment | Errors per home page | Source |
|---|---|---|
| .gov domains | 18.5 | WebAIM |
| Government category | 42.4 | WebAIM |
| Shopping | 71.0 | WebAIM |
| Sports | 71.4 | WebAIM |
| Astro sites | 9.0 | WebAIM |
| WordPress sites | 52.8 | WebAIM |
| Shopify sites | 75.1 | WebAIM |
A different lens tells a similar story. The Web Almanac 2025 accessibility chapter puts the median Lighthouse accessibility score at 85 (83 in 2019), with only 30% of pages meeting color-contrast checks and 67% removing focus outlines. Overlay widgets appear on 2.1% of desktop pages.
James Porter: More ARIA, more errors. ARIA is not a fix; it is a promise to the screen reader that you will behave like a native control. Use the native control. Then check contrast, alt text and labels. Those three cover most of the list above.
What Percentage of Websites Use HTTPS?
97.3% of mobile home pages and 97.5% of desktop home pages load over HTTPS, and 98.8% of mobile requests are encrypted (Web Almanac 2025 security chapter). Encryption is solved. Hardening is not.
| Security control (July 2025) | Pages | Source |
|---|---|---|
| HTTPS (mobile home pages) | 97.3% | HTTP Archive |
| TLS 1.3 | ~76% | HTTP Archive |
| Let’s Encrypt certificate (mobile) | 52.6% | HTTP Archive |
| HSTS header (mobile) | 36% | HTTP Archive |
| Bot protection service (mobile) | 31.1% (26.5% in 2024) | HTTP Archive |
| Content-Security-Policy (mobile) | 21.9% (18.5% in 2024) | HTTP Archive |
On Cloudflare’s network, post-quantum encryption protected 52% of human HTTPS traffic by early December 2025, up from 29% at the start of the year, while HTTP/3 carried 21% of requests (Cloudflare Radar 2025 Year in Review). Threat data sits on our DDoS attack statistics page.
James Porter: A certificate takes one cron job. A Content-Security-Policy takes an audit of every script you load, which is why four pages in five still skip it. Start in report-only mode. Read the reports for a week. Then enforce.
How Much Website Traffic Comes From Bots and AI Crawlers?
On December 2, 2025, humans generated 47% of HTML requests on Cloudflare’s network and non-AI bots 44%; AI bots averaged 4.2% over January 1 to December 2, 2025, peaking at 6.4% in late June. Googlebot alone averaged 4.5% (Cloudflare Radar 2025 Year in Review).
| Verified bot traffic, 2025 | Share | Source |
|---|---|---|
| Search engine crawlers (category) | 40% | Cloudflare |
| AI crawlers (category) | 20% | Cloudflare |
| Googlebot | >28% | Cloudflare |
| GPTBot | ~7.5% | Cloudflare |
| Bingbot | 6% | Cloudflare |
Do AI crawlers send traffic back?
Barely. Cloudflare’s crawl-to-refer ratio counts pages crawled per visitor referred back.
| Operator | Crawl-to-refer, 2025 | Source |
|---|---|---|
| Anthropic | peak 500,000:1, then ~25,000:1 to ~100,000:1 | Cloudflare |
| OpenAI | peak 3,700:1 (March) | Cloudflare |
| Perplexity | below 200:1 from September | Cloudflare |
| Microsoft | peaks of 50:1 to 70:1 | Cloudflare |
| 3:1 to 30:1 | Cloudflare |
Axis Intelligence Research calculates that Anthropic’s low end (25,000:1) was 833 times Google’s high end (30:1). GPTBot, ClaudeBot and CCBot were the user agents most often fully disallowed in robots.txt. The search-side consequence is covered in our zero-click search statistics and SEO statistics.
James Porter: A crawler that takes 25,000 pages and returns one visitor is a bandwidth bill, not a referral source. Decide per user agent. Allow the ones that send traffic, rate-limit the rest at the edge, and log what they fetch before you block anything.
The Web Vital Signs Index (WVSI)
The Web Vital Signs Index (WVSI) is Axis Intelligence Research’s 0–100 composite of the open web’s health. Its baseline reading is 42.2 (October 2026). It joins five primary datasets that are usually read in isolation: activity, speed, security, inclusion and human presence. Higher is healthier; each component is a share of sites, pages or requests, so it maps directly to 0–100.
| Component | What it measures | Input | As of | Weight | Source |
|---|---|---|---|---|---|
| Pulse | Active share of all websites | 14.5 | Jul 2026 | 20% | Axis calc. from Netcraft |
| Speed | Mobile origins with good Core Web Vitals | 48.0 | Jul 2025 | 20% | HTTP Archive |
| Security | Mobile home pages served over HTTPS | 97.3 | Jul 2025 | 20% | HTTP Archive |
| Inclusion | Top-1M home pages with no detected WCAG failures | 4.1 | Feb 2026 | 20% | WebAIM |
| Human presence | HTML requests from humans | 47.0 | Dec 2, 2025 | 20% | Cloudflare |
WVSI = 0.20 × (14.5 + 48.0 + 97.3 + 4.1 + 47.0) = 42.2
Equal weights keep the index transparent and reproducible from the CSV. Each input carries its own as-of date, so the WVSI is a dated snapshot that moves when any source publishes. The spread is the finding: one component near 100, two below 15. The web has encrypted itself almost completely and done close to nothing about who can use it.
James Porter: Security scores 97 because browsers forced it and Let’s Encrypt made it free. Inclusion scores 4 because nothing forces it yet. The pattern says defaults move the web, not best-practice guides. Put accessibility checks in CI, where they block a merge, and the number moves.
Methodology
Collection. Every figure was retrieved on October 18, 2026 from the primary publisher: Netcraft Web Server Surveys (October 2024; January to July 2026), the Verisign Domain Name Industry Brief Q2 2026, the HTTP Archive Web Almanac 2025 (Performance, Page Weight, Accessibility and Security chapters; July 2025 crawl of 17.2 million websites), The WebAIM Million 2026 (February 2026, top one million home pages) and the Cloudflare Radar 2025 Year in Review (January 1 to December 2, 2025). No figure comes from a secondary compilation.
Formulas. Active sites = vendor active-site count ÷ vendor active share, using Cloudflare because it has the largest active count. Non-active sites = total sites − active estimate. Growth = (end ÷ start − 1) × 100. Dormant-per-active ratio = change in non-active sites ÷ change in active sites. Responding-domain share = Netcraft domains ÷ Verisign registrations. WVSI = equal-weighted mean of five 0–100 inputs. Each calculated row in the CSV states its inputs in method_note.
Scope notes. Netcraft counts hostnames and its definition of activity excludes generated and duplicate content; the estimate inherits that definition. Web Almanac and Chrome UX Report cover origins with enough Chrome traffic; WebAIM covers home pages only and automated checks only; Cloudflare data reflects traffic on its own network. Source discrepancies found during verification are reported where they occur.
About this dataset
- File:
website-statistics.csv, 315 rows, long format, UTF-8. - Columns: metric, value, unit, as_of_date, dimensions, source_org, source_document, source_url, retrieved_date, is_primary, axis_calculated, method_note.
- License: CC BY 4.0. Reuse freely with attribution to Axis Intelligence Research.
- Mirrors: Hugging Face, Kaggle and GitHub under Axis Intelligence Research [insert repository links after upload].
- Cite as: Axis Intelligence Research. (2026). Website Statistics 2026 [Data set]. Axis Intelligence.
Cite this page
APA Axis Intelligence Research. (2026, October 18). Website statistics 2026: How many websites exist, how many are alive, and the web’s vital signs. Axis Intelligence. https://axis-intelligence.com/website-statistics/
MLA Axis Intelligence Research. “Website Statistics 2026: How Many Websites Exist, How Many Are Alive, and the Web’s Vital Signs.” Axis Intelligence, 18 Oct. 2026, axis-intelligence.com/website-statistics/.
Chicago Axis Intelligence Research. “Website Statistics 2026: How Many Websites Exist, How Many Are Alive, and the Web’s Vital Signs.” Axis Intelligence, October 18, 2026. https://axis-intelligence.com/website-statistics/.
FAQ: Website Statistics 2026
Why do different sources say there are 1.1 billion, 1.5 billion or 200 million websites?
They count different things. Netcraft’s 1.49 billion counts hostnames that answer HTTP. Its domain count is 305.3 million. Axis Intelligence Research’s active-site estimate is 216.7 million, which excludes parked and auto-generated pages. Older figures usually reflect earlier Netcraft surveys: 1.13 billion was the October 2024 count.
Is the number of websites still growing in 2026?
Yes, by every measure, but unevenly. From January to July 2026 hostnames rose 8.6%, domains 3.5%, web-facing computers 2.8% and active sites 4.4%. Domain registrations grew 8.1% year over year to 401.6 million.
How many new websites are created each month?
Netcraft’s July 2026 survey recorded 5.5 million more sites than June. Between October 2024 and July 2026 the hostname count grew by about 364 million, roughly 17 million a month on average, but only about 1.1 million a month of that growth was active sites.
Does failing Core Web Vitals mean a site is slow for most users?
No. A site passes only if its 75th-percentile visit meets the LCP, INP and CLS thresholds together. Failing one metric fails the set. On mobile, LCP is the usual culprit: 62% of origins pass it, versus 77% for INP and 81% for CLS.
Which sites are most and least accessible?
In The WebAIM Million 2026, .gov home pages averaged 18.5 detected errors and Astro-built sites 9.0, while shopping (71.0) and sports (71.4) sites averaged the most. Using ARIA attributes correlated with more errors, not fewer.
Should I block AI crawlers in robots.txt?
It depends on what a crawler returns. Cloudflare measured Google sending one visit per 3 to 30 crawls in 2025, against one per 25,000 to 100,000 for Anthropic after its peak. Blocking stops content use for training only for crawlers that honor robots.txt; it also removes any referral those crawlers might send.
Is HTTPS still a differentiator for a website?
No. 97.3% of mobile home pages already use HTTPS. The gaps are headers: only 36% of mobile pages send HSTS and 21.9% send a Content-Security-Policy.
What does a WVSI of 42.2 mean, and will it change?
It means the average of five health signals sits below the midpoint, because activity (14.5) and accessibility (4.1) are low despite near-universal HTTPS (97.3). It changes whenever one of the five sources publishes a new edition; each input’s date is listed in the dataset.