SEO Statistics 2026: What 6,767 Real Homepages Show

What do the SEO statistics for 2026 show?

Share of scanned homepages and sites with each gap (November 2026 sweep)

Category	Value
No JSON-LD structured data	54%
No XML sitemap found	32%
No meta description	22%
No llms.txt file	82%
Source: KinetixSEO, 2026-11-12 (https://kinetixseo.com/learn/methods/research-sweep/2026-11-05-tranco-JZPGY-5e213359)

Across the homepages KinetixSEO scanned in November 2026, 54% of 6,586 carry no JSON-LD structured data, no XML sitemap could be found for 32% of 6,496, 22% of 6,587 have no meta description and 82% of 6,226 publish no llms.txt file. All figures come from the research sweep edition of 12 November 2026, a seeded random draw of 10,000 popular domains.

What we checked	Share of sites	Sites it rests on
No JSON-LD structured data on the homepage	54%	6,586
No XML sitemap found	32%	6,496
No meta description	22%	6,587
At least one image without alt text (sites with images)	21%	5,203
Not all three of privacy policy, contact page and about page linked	68%	6,587
No llms.txt file	82%	6,226
Block at least one AI crawler in robots.txt (minimum)	14%	6,574

Every share has its own number of sites behind it. A check only counts the sites where it could actually run, so a homepage whose sitemap request timed out is left out of the sitemap share instead of being counted as "no sitemap".

How was this measured?

One homepage per site, one attempt each. I drew 10,000 domains out of the 99,001 ranked 1,000 to 100,000 on Tranco list JZPGY, a research ranking of popular domains. The draw uses a public seed, and the list, the seed and its sha256 hash are on the edition page, so anyone can redraw exactly the same sample. The scans ran in November 2026 with one unchanged version of the scanner, and the full aggregate results are downloadable there as CSV and JSON.

Know what this is before you quote it. These are homepages of popular domains. Not every page of those sites, and not the whole web. The very top of the list is left out on purpose, because that band is mostly the big platforms, and so is the long tail, where the list fills up with parked domains and redirects. What is left is the broad middle of popular domains. It still holds service domains with no homepage at all, which is part of why so many drawn domains could not be scanned.

What happened to the 10,000 sites?

Of 10,000 domains drawn, 6,767 homepages could be scanned. 2,781 failed on the site's side, 359 blocked our crawler in robots.txt, 33 answered with something that was not an HTML page and 60 failed on our side and were left out, according to the research sweep.

This part surprised me the most. 1,451 of the 10,000 sites, 14.5%, turned the crawler away before it saw a single page: 1,092 with an HTTP 403, 406 or 429 or a bot-challenge page, and 359 through robots.txt. Our crawler says who it is and asks politely, once.

A lot of popular sites do not want to talk to a bot they do not know.

That is fine for scrapers. It is less fine if the same rules also stop a crawler you do want, like the ones that feed AI answers. Many firewalls let verified crawlers such as Googlebot through, so this does not mean these sites block Google. It does mean a lot of sites run bot rules nobody looked at in a while. The Bot View Checker shows what your own firewall does with an unknown bot.

How many websites use structured data?

54% of 6,586 scanned homepages carry no JSON-LD structured data, so 46% carry some, per the research sweep. Structured data is a block of JSON in the page that tells search engines and AI systems what the page is: an organization, a product, an article, a local business. Without it they have to work that out from the text. In search it also makes a page eligible for rich results.

Only JSON-LD was counted. A site that writes its markup as Microdata or RDFa counts as "none" here, so the real share of sites with any structured data is a bit higher. Still. More than half of popular homepages do not say in machine-readable form who they are. The practical JSON-LD guide shows what to add first, the JSON-LD statistic page keeps this number current with every new edition, and the Structured Data Validator shows what your own page declares.

How many websites have no XML sitemap?

No XML sitemap could be found for 32% of 6,496 scanned sites, per the research sweep. The scanner looks for a sitemap declared in robots.txt and at the usual paths (/sitemap.xml, /sitemap_index.xml, /wp-sitemap.xml). A sitemap kept somewhere unusual and declared nowhere counts as missing, which is also what a crawler looking in the same places would conclude.

Does a small site need one? Not strictly, a small and well-linked site gets found anyway. A bigger site, or one that adds pages often, waits longer for new pages to be crawled without one. It is a quick fix with almost any CMS. See how many websites have no XML sitemap for the running figure, and check your own with the Robots & Sitemap Validator.

How many websites have no meta description?

22% of 6,587 scanned homepages have no meta description, per the research sweep. Without one, Google writes its own snippet from the page text, which on a homepage is often a cookie banner or a menu line. Google rewrites plenty of descriptions anyway. But a written one is the only part of the search snippet you get to suggest, and on a homepage that is the first sentence a stranger reads about your company.

The on-page SEO checklist covers how to write one, and the Meta Tag Checker shows the title and description your page declares.

What else is missing on the average homepage?

The edition's downloadable data has every check the scanner runs, counted over all 6,767 scanned homepages. Three stood out to me:

• 30% have no H1 heading. Google does not need an H1 to rank a page, but it often uses the main heading when it rewrites the title in its results, and screen-reader users navigate by it.
• 37% declare no canonical tag. On a homepage that is rarely a disaster, but it leaves search engines to pick between the www and non-www, http and https, and tracking-parameter versions by themselves. More on that in canonical URLs.
• 4.1% carry a noindex directive. That is 280 homepages of popular domains telling Google not to show them in search at all. Some do it on purpose. I doubt all of them do.

These come from the full data download, where a check that could not run on a site counts as not firing, so each share is a minimum.

How many homepages skip alt text or a trust page?

On 21% of 5,203 homepages that show images, at least one image has no alt text, per the research sweep. One missing alt is enough to count, so this says how common the slip is, not how bad each site is. An image without alt text gives screen-reader users no description and gives image search less to go on.

68% of 6,587 scanned homepages do not link all three of a privacy policy, a contact page and an about page. Only links on the homepage itself were read, so a site that tucks its about page in an inner menu counts as missing it. These are the pages that tell a visitor, a search engine and an AI model who runs a site and how to reach them. The Trust Signals Checker checks your own homepage for them.

How many websites have llms.txt or block AI crawlers?

82% of 6,226 scanned sites publish no llms.txt file, so 18% publish one, per the research sweep. A site whose /llms.txt request got no usable answer (a timeout, a server error, a firewall refusal) is left out instead of counted as having none. A site that answered with an HTML page in place of the file counts as having none. Whether llms.txt changes how often AI answers cite a site is not proven, as what llms.txt is and whether you need one explains. I track it because people ask about it, not because I think it is the fix.

14% of 6,574 scanned sites block at least one AI crawler in robots.txt. That is a minimum: a robots.txt we could not read for another reason counts as "not blocking". Blocking a training crawler such as GPTBot keeps your text out of model training. Blocking a search crawler such as OAI-SearchBot or PerplexityBot tells that engine not to fetch your pages for its answers, which is a different decision and one a lot of sites make by accident. The AI Crawler Checker shows per bot what your robots.txt allows. Running numbers: how many websites block AI crawlers and how many websites have llms.txt.

Does HTTPS still make a difference?

Almost everyone has it now. Only 111 of the 6,767 scanned homepages were not served over HTTPS, per the research sweep. Those 111 had a median KinetixSEO SEO score of 49, against 70 for the 6,656 served over HTTPS. That is our own score, and it is a correlation, not proof that HTTPS alone moves anything. A site that skipped HTTPS in 2026 usually skipped a lot of other things too, and that is probably the real story.

Did anything change since October?

No. And that is the honest answer.

The October edition scanned 3,096 homepages from a different random draw. Its shares sit within a few points of these: 52% without JSON-LD then, 54% now; 20% without a meta description then, 22% now. Each edition is a new draw of different sites, and the October scanner had gaps that the November one fixed: about 350 sites whose robots.txt it could not read were never scanned, and those lean towards neglected sites, so October read slightly better. A few points between two editions is sampling noise and a better scanner, not the web getting worse in a month. The October write-up, What 5,000 Homepages Reveal, stays up as the record of that edition. Real trends need more editions, and they will come.

How do you check your own website?

Run the free SEO and GEO check on your homepage. It is the same scanner that ran this sweep, on one site, without an account. It shows each of the checks above for your site: structured data, sitemap, meta description, H1, canonical, alt text, trust pages, llms.txt and AI crawler rules.

What these SEO statistics mean for your site

Most popular websites do not lose on some clever AI trick. They lose on the basics. A homepage that does not say who it is in structured data, has no sitemap, no meta description and no about page is the normal case, not the exception. Fix those first. The AI-specific files come after.

Lesson learned: the bar is lower than you think. That is good news if you clear it :-)

Frequently asked questions

What are the most important SEO statistics for 2026?

In KinetixSEO's November 2026 research sweep, 54% of 6,586 scanned homepages carry no JSON-LD structured data, no XML sitemap could be found for 32% of 6,496 sites, 22% of 6,587 have no meta description and 68% of 6,587 do not link a privacy policy, a contact page and an about page.

How many websites have structured data in 2026?

46% of the 6,586 homepages scanned in November 2026 carry JSON-LD structured data, so 54% carry none. Markup written as Microdata or RDFa was not counted, so the share with structured data in any form is somewhat higher.

How many websites block AI crawlers?

At least 14% of the 6,574 sites scanned in November 2026 block at least one AI crawler in robots.txt. Separately, 14.5% of the 10,000 drawn sites turned away a declared research crawler before serving the page, mostly through firewall or bot-protection rules.

How many websites have an llms.txt file?

18% of the 6,226 sites that answered the request in November 2026 publish an llms.txt file, and 82% have none. No AI company has confirmed its crawler uses the file.

Where can I download the data?

The full aggregate results are on the edition page as CSV and JSON under a Creative Commons Attribution license, together with the domain list, the seed and its sha256 hash.