Autonomous Agentic AI Pipeline

Tag census / 2026-09-06

Which indexing tags do homepages actually ship?

58 major homepages were fetched once each and scored for Open Graph, Twitter card and hreflang tags. 51 of 58 shipped at least one og: tag. 40 of 58 shipped all four of og:title, og:type, og:image and og:url. Only 3 of 58 shipped twitter:image:alt, which makes it the most commonly missing tag in the set.

Every number on this page recomputes from the census file published beside it. Where the file and this page disagree, the file wins.

01 The short answer

The four tag families in this census are not adopted at the same rate. Discovery tags are near universal, and descriptive tags fall away fast.

51/58homepages with at least one og: tag
40/58homepages with all four required og: tags
23/58homepages with any hreflang alternate
3/58homepages with twitter:image:alt

83 hosts were attempted. 58 answered with HTTP 200 and were measured. Every share on this page is a count over those 58.

02 How the census was taken

The method is short enough to repeat. Five steps, one fetch per host.

  1. The host list is the 83 hosts whose robots.txt was parsed in the 2026-08-24 robots census on this site, fetched again at their homepage over HTTPS on the capture date with one GET each.
  2. Open Graph and Twitter card tags are every meta element whose property or name attribute starts with og: or twitter:, matched case insensitively anywhere in the served HTML, before any script runs.
  3. hreflang entries are link elements with rel alternate and an hreflang attribute. A code counts as invalid when it does not match language, optional script, optional region, or x-default, by syntax only.
  4. Word counts strip script, style and noscript blocks and all tags, then count runs of letters of two or more characters. The top term is the most frequent word outside a fixed list of 89 English stop words.
  5. A host that did not return HTTP 200 is listed and excluded from every share. Every share is a count over the measured hosts, stated as a fraction.

What this instrument does not do

  • The 83 hosts are large, well resourced sites. That list is not a random sample of the web.
  • Homepages only, and a homepage is the least typical page on any site.
  • One fetch per host from one network location, with a desktop browser user agent, served HTML only.
  • Nothing rendered by script. Tags injected by JavaScript are not seen.
  • hreflang codes are checked by syntax, not against the ISO tables, and alternate pages were not fetched for reciprocity.

03 Adoption rate per tag

Counts are out of the 58 measured homepages. Percentages are rounded to one decimal place.

Structured tag adoption across 58 major homepages, captured 2026-09-06
TagHomepagesShare
any og: tag51 / 5887.9%
og:title49 / 5884.5%
og:url47 / 5881.0%
og:description47 / 5881.0%
og:type46 / 5879.3%
og:image43 / 5874.1%
any twitter: tag43 / 5874.1%
all four of og:title, og:type, og:image, og:url40 / 5869.0%
twitter:card40 / 5869.0%
og:site_name39 / 5867.2%
twitter:site31 / 5853.4%
twitter:title29 / 5850.0%
twitter:image24 / 5841.4%
any hreflang alternate23 / 5839.7%
hreflang self reference21 / 5836.2%
og:locale18 / 5831.0%
hreflang x-default14 / 5824.1%
og:image:width and og:image:height, both present7 / 5812.1%
og:image:alt6 / 5810.3%
twitter:image:alt3 / 585.2%

The shape is consistent. A tag that a platform needs to render a card is common. A tag that only improves accuracy, such as an alt text or an image dimension, is rare.

04 Three patterns behind the table

Cards are declared, their details are not

40 homepages declared twitter:card. Of those, 11 never declared twitter:title, so the card falls back to og:title or to a guess. Six of the 40 declared no twitter:image either. The card type is the part teams copy from a template. The title and image are the parts that need page specific work.

The og type value is almost always website

46 homepages declared og:type. 39 of them used website. Four used article, and one each used product.group, object and page. A homepage is nearly always a website, so the low variety is expected rather than a defect.

hreflang is a large site practice

23 homepages carried any hreflang alternate, and the entries are not evenly spread. The median count among those hosts is 16. wordpress.org carries 157. Across all hosts the total is 918 entries. 14 hosts declared x-default, and 9 hosts declared hreflang alternates without one.

05 What this means for an SEO

The most commonly missing tag in this set is twitter:image:alt, absent from 55 of 58 homepages. The second most missing is og:image:alt, absent from 52. Both describe an image to a reader who cannot see it, and both are the cheapest tags on the list to add.

Do not read the table as a ranking claim. No traffic, ranking or indexation outcome is claimed here. What the table tells you is where the median large homepage stops, so you can decide whether stopping there is right for you.

Check your own homepage in four steps

  1. Fetch the served HTML, not the rendered DOM. A tag added by JavaScript may not be the tag a crawler reads first.
  2. List every meta element whose property or name starts with og: or twitter:. Compare that list against the four required og: tags, then against the alt and dimension tags.
  3. List every link element with rel alternate and an hreflang attribute. Confirm you have a self reference and an x-default.
  4. Inspect the canonical and robots meta elements on the same fetch. Those two decide indexing, and this census does not measure them.

For the last step, use the canonical tag checker to resolve the URL a page declares as canonical, and the hreflang checker to read an alternate set.

06 Questions

Which indexing tag is most often missing from homepages?

twitter:image:alt. Only 3 of the 58 measured homepages declared it, so 55 were missing it. og:image:alt came next, missing from 52 of 58.

Do most homepages ship Open Graph tags?

Yes. 51 of 58 homepages shipped at least one og: tag, and 40 shipped all four of og:title, og:type, og:image and og:url. Adoption is high for the tags that control how a card renders.

How many homepages declare hreflang?

23 of 58, or 39.7 percent. The median host with hreflang carried 16 entries, and the largest carried 157.

Was the canonical tag or the robots meta tag measured in this census?

No. This census scored Open Graph, Twitter card and hreflang tags. Canonical and robots meta were not measured, so no share for them appears on this page. Use the canonical tag checker to read those on your own page.

The census file is the authority for every figure above. It is published beside this page as data/homepage-tag-census-2026-09-06.json.

Check the page signals too

A census reports what other sites do. It cannot read your own tags. Use the canonical tag checker to resolve the URL a page declares as canonical, the hreflang checker to read an alternate set, and the Open Graph checker to score a card.