We couldn't find Organization or WebSite schema on the homepage

The check looks only at the home page (identified by its address: the site root, or the root of a language version) and asks whether it declares the type Organization or WebSite. Either one is enough. When neither is there, the finding is raised.

Since 19 September 2026 the check looks for that type in both formats we read: JSON-LD blocks and schema.org microdata (itemtype="https://schema.org/Organization"). The format does not decide it; the presence of the type does.

Types are also found inside @graph and inside arrays, so the usual pattern of one block holding several entities is read correctly.

What the check does not do:

• It does not judge what the schema says. An Organization with no name, no logo and no address passes just like a carefully filled one.
• It runs on no page other than the home page. Not on language variants, not on subpages.
• It does not check that the identity in the schema is real: whether the name matches the one in your header, or whether a social profile link leads to your profile.
• It cannot see RDFa, same as SCH-01. Microdata it has read since 19 September 2026; RDFa it does not. That is an admitted gap, not a claim that the format is invalid.
• It does not deal with structured data missing altogether. That is SCH-01, and when a page has microdata but no JSON-LD, SCH-06.

The finding attaches to the home page; severity is warning.

Everything from SCH-01 applies: the effect of structured data on whether a language model cites you is not documented. The measurement (Ahrefs, 1,885 pages against 4,000 controls) landed within noise, with a significant decline for AI Overviews, and the Otterly controlled experiment found that information placed only in schema was used by no platform.

Organization and WebSite do have one thing the other types do not, and it is fair to name it: they describe identity, not content. They answer “who runs this site”, and that is information you need stated clearly on the site anyway: for people, for credibility, and in the EU for the legal obligation to identify the operator.

The documented benefit is with Google, and only for classic search: it uses structured data to qualify pages for rich results. For WebSite with a search action and Organization with a logo, that is often visible directly in results. It is not something we would promise for AI, but it is not nothing either.

How to place this finding in your queue: it is the cheapest item of the whole SCH-* group: one block on one page, done in ten minutes. If you are going to do structured data at all, start here. If you are not going to do it at all, you forfeit nothing documented. Both are legitimate decisions, and the warning severity is there chiefly because Google recommends structured data itself, not because we measured a loss.

Add one block to the home page with both entities linked together. The example is JSON-LD, but if your structured data lives in microdata, adding these two types in that same format is enough (itemtype="https://schema.org/Organization"): Google treats the formats equally and so does this check. This is the whole foundation:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Organization",
      "@id": "https://your-domain/#organization",
      "name": "Our Company Ltd",
      "url": "https://your-domain/",
      "logo": "https://your-domain/logo.png",
      "sameAs": [
        "https://www.linkedin.com/company/our-company",
        "https://www.facebook.com/ourcompany"
      ]
    },
    {
      "@type": "WebSite",
      "@id": "https://your-domain/#website",
      "url": "https://your-domain/",
      "name": "Our Company",
      "publisher": { "@id": "https://your-domain/#organization" }
    }
  ]
}
</script>

What is worth noticing in that snippet:

• @id and the reference to it. Instead of describing the Organization twice, WebSite just points at it. Use the same trick on subpages: an article refers to #organization rather than repeating the name and logo.
• sameAs is the most useful field in the block. These are links to your verifiable profiles elsewhere: LinkedIn, Wikipedia, a companies register, a trade directory. They are what connects your site to an entity that can be checked independently of you.
• logo must be an absolute URL to a publicly reachable image.
• The name must match what is visible on the site. A trading name in the schema and a different one in the header is a contradiction that helps nobody.

Two things worth doing at the same time, even though this check does not look at them:

1. Write the same details into the text. An “About us” page with the name, address and registration number does more for credibility than schema, and it is information reachable even by a model that reads no structured data at all.
2. Reuse Organization on subpages as the publisher of articles (SCH-01). It is one extra line and it keeps identity consistent across the site.