The Schema Types That Still Earn Something in 2026

Google's structured data gallery listed 25 supported features when we counted it on 15 August 2026. FAQ is not one of them. Neither is HowTo. Both used to be, and both were retired without the markup becoming invalid.

That gap is where the arguments come from. "Does this schema type still matter" is two questions wearing one coat. Does Google draw a visual feature from it? And can a machine still read the fact out of it? The answers differ by type, and the second one has barely moved since 2020.

The short answer

• Google's rich results gallery lists 25 features. Article, Breadcrumb, Product, Organization, Recipe, Video and Q&A are there. FAQ and HowTo are not.
• Google's own AI guidance says structured data "isn't required for generative AI search, and there's no special schema.org markup you need to add".
• Retired markup is not invalid markup. Google has said since 2023 that unused structured data causes no problems for Search.
• Our own homepage carries no JSON-LD at all, which we checked while writing this and have not fixed yet.

What Google still shows a rich result for

The gallery is the only list that counts, because it is the list Google maintains. Everything else is someone's memory of what used to work. Here are the entries a normal content or SaaS site could plausibly earn, with what each one actually does.

Note the last two. Q&A survived the cull that killed FAQ, and they look similar enough that people conflate them. The difference is who wrote the answer. QAPage is for a page where a user asked a question and other users answered it. FAQPage was for a page where the site owner wrote both halves. Google kept the one a site owner cannot manufacture alone and dropped the one anybody could write in an afternoon.

Breadcrumb also carries a limit worth knowing. Google added a feature availability note stating the feature appears on desktop only, because breadcrumbs truncate badly on small screens. So roughly half your traffic never sees the thing you marked up.

What got retired, and when

Two retirements matter for content sites, and Google logged both in its own changelog.

HowTo went first. A Search Central post from 8 August 2023 limited HowTo rich results to desktop, then a 14 September update finished the job: "As of September 13, Google Search no longer shows How-to rich results on desktop, which means this result type is now deprecated." The documentation was removed in the same month.

FAQ took longer. The documentation changelog entry for 8 May 2026 reads: "Added a deprecation notice to the FAQ rich result documentation. Why: This feature will no longer appear in Google Search starting May 7, 2026." On 15 June 2026 the doc itself went: "Removed documentation for the FAQ rich result feature."

Here is the part you can check yourself in ten seconds. Both documentation URLs still resolve, and both now redirect to the changelog entry announcing their own removal. /search/docs/appearance/structured-data/faqpage sends you to /search/updates#removing-faq-rich-result. /search/docs/appearance/structured-data/how-to sends you to /search/updates#how-to-deprecation. Each page now points at its own obituary.

A rich result and a readable fact are different things

Google's guide to optimizing for generative AI features has a section on mistakes, and one of them is overfocusing on structured data. It reads: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add. However, it's a good idea to continue using it as part of your overall SEO strategy, as it helps with being eligible for rich results on Google Search."

There is no AI-specific schema to buy. The same guide tells you to ignore chunking your content and creating AI text files like llms.txt. It also names AEO and GEO directly, as labels for work that is still SEO. We covered what llms.txt actually is and who reads it and the difference between SEO, AEO and GEO at length.

So what does retired markup still do? It stays a valid, parseable statement of fact in a standard format, on a page that anything can fetch. schema.org still defines FAQPage and HowTo with no deprecation notice, because schema.org is a vocabulary and Google is one consumer of it. Nobody has published a measurement showing that FAQPage markup changes how often an answer engine cites you. We have not measured it either, and we are not going to claim it.

What we can say is that deleting it buys you nothing. Google's 2023 advice on exactly this situation still reads: "While you can drop this structured data from your site, there's no need to proactively remove it. Structured data that's not being used does not cause problems for Search, but also has no visible effects in Google Search."

Three real pages, checked today

We built a schema validator that encodes Google's documented required and recommended fields, then pointed it at pages we did not control. All three results below are from 15 August 2026 and you can reproduce them.

web.dev/articles/lcp, a Google property, ships this:

{
  "@context": "https://schema.org",
  "@type": "Article",
  "dateModified": "2025-09-04",
  "headline": "Largest Contentful Paint (LCP)"
}

Two properties. No image, no datePublished, no author. Every one of those is on Google's own recommended list for Article. The markup is still valid, because Article has no required properties at all, but Google is not following its own recommendations on a Google site.

developer.mozilla.org/en-US/docs/Web/HTTP returned zero JSON-LD blocks and five RDFa elements. Mozilla's reference documentation for HTTP carries no JSON-LD on that page at all. Whatever reads it reads the HTML, which is what a parser was always going to do anyway.

gov.uk/vat-rates carries a complete FAQPage with two questions and a BreadcrumbList, three months after the rich result died. Government and health sites were the last category still eligible for the FAQ result before it went, so this markup outlived its own feature. Nobody at GDS has taken it down, and by Google's stated position, nobody needs to.

The four types worth the effort on a content site

If you publish articles and sell software, this is the whole list. One JSON-LD block, one @graph, four nodes that reference each other.

{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "BlogPosting",
      "@id": "https://example.com/blog/post#article",
      "headline": "The Schema Types That Still Earn Something in 2026",
      "image": "https://example.com/blog/post/og.png",
      "datePublished": "2026-08-15T09:00:00Z",
      "dateModified": "2026-08-15T09:00:00Z",
      "author": { "@id": "https://example.com/#author" },
      "publisher": { "@id": "https://example.com/#org" }
    },
    {
      "@type": "Person",
      "@id": "https://example.com/#author",
      "name": "Adarsh Mishra",
      "url": "https://example.com/about"
    },
    {
      "@type": "Organization",
      "@id": "https://example.com/#org",
      "name": "Example",
      "url": "https://example.com",
      "logo": "https://example.com/icon.png"
    },
    {
      "@type": "BreadcrumbList",
      "itemListElement": [
        { "@type": "ListItem", "position": 1, "name": "Blog", "item": "https://example.com/blog" },
        { "@type": "ListItem", "position": 2, "name": "The Schema Types That Still Earn Something in 2026" }
      ]
    }
  ]
}

Two details in there are documented and routinely got wrong. The last breadcrumb has no item. Google's doc says item is not required on the last crumb, and Google falls back to the containing page's URL. And author is a reference to a Person node carrying a url, not a bare string. Google recommends author.url as a link to a page that uniquely identifies the author.

Selling something? Product needs name plus one of offers, review or aggregateRating. Inside an Offer, price or priceSpecification.price is required and priceCurrency is recommended for product snippets but required for merchant listings, so provide it either way. Inside an AggregateRating, you need ratingValue and at least one of ratingCount or reviewCount. If you omit bestRating, Google assumes 5.

The generators for each of these are free and need no signup: article, breadcrumb, product.

Our own homepage has no JSON-LD

We checked while writing this. https://seobuilder.tech/ returns zero application/ld+json blocks. Our blog posts carry a full @graph and our tool pages carry markup. The homepage carries nothing, and the homepage is the page an answer engine would resolve "what is SEOBuilder" against.

For a company selling AI visibility measurement, that is embarrassing. Nobody decided to skip it. The blog shipped with schema because schema was part of the blog work, and the homepage predates that decision. It is on the list now, dated, at the same priority as anything else that is broken.

For the same check on your own pages, structured data for AI reports whether the nodes form one connected graph or a pile of disconnected fragments. That is the failure mode that survives a green Rich Results Test.

What to do next

Run one page through a validator and read what comes back rather than assuming. If Article is there and author is a bare string, fix that first, because author identity is the property Google writes the most guidance about. If you have FAQ or HowTo markup, leave it where it is and go do something that pays.

Then measure the part schema does not control. Whether Google's AI Overview names you for the questions your buyers ask is a different question from whether your markup validates. The two do not track each other. SEOBuilder, our product, scans live AI answers and returns the citation URLs, so you can see who got named instead of you. To be exact about scope, that is seven engines, Google AI Overviews and ChatGPT included. No third party has access to Google's internal AI systems, and Google says so in the same guide that tells you to skip the schema hacks.

If the AI question is the one you are stuck on, start with how to measure AI visibility, then what the tools in this category do and cost.