GEO
We Checked 62 Big-Site Homepages: Who Publishes Entity Schema and Trust Pages?
Published 2026-11-04
Two checks in an AI-readiness audit ask who is behind a site: structured data that names the organization, and trust pages like About, Contact and Privacy. We wanted to know how often well-known sites actually publish them, so on 2026-11-04 we read the homepage HTML of 73 sites and counted.
How we measured
We picked 73 well-known sites across news, SaaS, e-commerce, reference, SEO tools and developer infrastructure. A script fetched each homepage with a desktop Chrome user agent on 2026-11-04. 62 returned readable HTML; 11 did not (nytimes.com, forbes.com, medium.com, etsy.com, ebay.com, tripadvisor.com, yelp.com and bloomberg.com answered 403, amazon.com answered 202, and washingtonpost.com and bestbuy.com timed out). Those 11 are excluded from every count below.
For each readable page the script parsed every application/ld+json block, including nodes inside an @graph. It counted a site as having entity schema when a node was an Organization, Corporation, NewsMediaOrganization, LocalBusiness, Person or similar. It counted sameAs links only on those entity nodes. For trust pages it looked for links whose address or text pointed to About, Contact or Privacy.
Entity schema: most sites have it, many giants do not
44 of 62 homepages (71%) had any JSON-LD, and 40 (65%) had entity schema. The most common types were:
| Type on the homepage | Sites (of 62) |
|---|---|
| Entity schema (Organization, Corporation and similar) | 40 |
| WebSite | 28 |
| WebPage | 19 |
| BreadcrumbList | 8 |
| SoftwareApplication | 8 |
| FAQPage | 3 |
The 18 homepages with no JSON-LD in their raw HTML include some of the best-known names on the list: github.com, shopify.com, slack.com, linkedin.com, microsoft.com, openai.com, anthropic.com, wikipedia.org and reddit.com. Four more had JSON-LD but no entity node (target.com, airbnb.com, healthline.com and nike.com). Brands that large are recognized without help, so the absence of markup tells you little about whether markup is useful. Also, none of the 62 homepages used Article-type schema, which belongs on article pages, not homepages.
sameAs: when a site links profiles, it links several
37 of 62 sites (60%) used sameAs on their entity schema, and none listed just one profile: the smallest list had two, the median was 6 and the largest was 12 (stripe.com). Three sites declared an organization but left sameAs out: wired.com, salesforce.com and walmart.com. The property's job is to tie a site to its official profiles so a system can recognize them as one entity (see the Organization schema generator for the markup).
Trust pages: Privacy first, Contact last
- Privacy: linked on 57 of 62 homepages
- About: 52 of 62
- Contact: 49 of 62
- All three: 45 of 62 (73%)
Four sites showed none of the three in their raw HTML: airbnb.com, anthropic.com, reddit.com and datadoghq.com. We would not read that as the pages not existing. Footers rendered by JavaScript are invisible to this method, which is also how many AI crawlers read a page, so for a crawler that does not run scripts the result is the same.
Putting the three together
30 of 62 sites (48%) had all of it on the homepage: entity schema, at least two sameAs profiles and links to About, Contact and Privacy. The other half were missing at least one piece. FAQPage structured data was rare at 3 of 62 (wix.com, squarespace.com and coursera.org).
What this does and does not tell you
- It shows what 62 hand-picked, mostly large US-based brands publish. It is not a random sample of the web.
- It reads only the raw homepage HTML. Schema on inner pages, or added by JavaScript, is not counted.
- Trust-page detection is a heuristic on link addresses and text, so unusual footers can be missed.
- It says nothing about whether any of this makes an AI system more likely to cite a site. We know of no vendor that has said it does.
The case for doing it anyway is practical: a small site has no brand recognition to lean on, and clear signals about who runs it cost little. HellSEO scores exactly these three things in the GEO section of an audit: Structured Citable Content, Entity Connection (sameAs) and Trust Pages. When we added sameAs and trust links to our own site, our GEO score rose from 70 to 85 (see our self-audit). That is one site, so treat it as an example rather than evidence. For the crawler side of the same question, see our 70-site crawler study.
Sources
Official documentation this post relies on (links checked 2026-11-04).
Frequently Asked Questions
How many big sites publish Organization schema on their homepage?
40 of the 62 homepages we could read (65%) carried entity schema in their raw HTML: Organization, Corporation, NewsMediaOrganization or similar. 35 of those used the plain Organization type. 18 homepages had no JSON-LD at all.
How many link their social profiles with sameAs?
37 of 62 (60%). Every site that used sameAs listed at least two profiles, with a median of 6 and a maximum of 12 (stripe.com). Three sites had entity schema but no sameAs: wired.com, salesforce.com and walmart.com.
Do big sites link About, Contact and Privacy pages?
45 of 62 (73%) linked all three from the homepage HTML. Privacy was the most common (57), then About (52) and Contact (49). A few sites that lacked a link, such as reddit.com or airbnb.com, may render their footers with JavaScript, which this method does not see.
Does this prove that schema and trust pages make AI cite a site?
No. It is a snapshot of what 62 well-known homepages publish, not a test of what AI systems cite. Large brands are recognized without any markup, so their choices say little about what a small, unknown site needs.
Related Free Tools
Related Posts
We Checked 70 Major Sites: Who Blocks AI Crawlers in 2026?
SEO vs AEO vs GEO: What's Actually Different (With Examples)
We Audited Our Own Site With HellSEO — Here's What We Found