The schema markup guide to rich results in 2026

If you're building a schema markup guide worth reading in 2026, start with a narrower question than "should I add schema." Only a specific, changing list of structured data types actually earns a Google rich result right now: Article, Product, Review snippet, Event, Job posting, Local business, Recipe, Video, Breadcrumb, Organization and Software app, among a few others. FAQPage and HowTo aren't on that list. Google stopped showing the FAQ rich result entirely as of May 7, 2026.

What it is and why it matters

Schema markup is structured data: a standardized, machine-readable label attached to page content so a search engine or an AI system can classify what that content actually is, rather than guessing from prose. It doesn't change what a visitor sees. It changes what a program reading the page understands.

Two things get conflated here. Schema.org is the full vocabulary, a collaborative project founded by Google, Microsoft, Yahoo and Yandex, now used across more than 45 million web domains with over 450 billion schema.org objects in place, according to schema.org's own overview. Google's rich-result support is a much smaller subset of that vocabulary. Every rich-result type is a schema.org type, but most schema.org types never produce a rich result at all.

Google recommends JSON-LD for implementation, over the two alternatives, Microdata and RDFa, calling it the easiest format for a site to implement and maintain at scale. All three are technically valid, but JSON-LD's script-tag structure keeps it separate from the visible markup, which is why most teams default to it. It travels well too: generate the block once and drop it into a page template without touching the surrounding HTML, useful on a site with hundreds of templated pages built from the same layout.

The commercial case for getting this right isn't abstract. Google's own documentation cites Nestle's measurement that pages appearing as rich results saw an 82% higher click through rate than pages that didn't. A rich result is a bigger, more specific piece of real estate in the results page, and it converts attention into clicks at a materially higher rate.

How it works in practice

The current Google rich result gallery

As of this schema markup guide's publish date, Google's supported structured data gallery lists roughly 25 eligible types. The commerce-facing ones, Product, Review snippet, Local business, show price, availability and rating information directly in the results page. Content-facing types like Article, Recipe, Video and Course list drive carousels and enhanced previews. A smaller utility group, Breadcrumb, Organization, Software app, Image metadata, adds context rather than a headline feature. If your structured data isn't producing a rich result, check whether the type you're using is still on this list at all, a step worth building into any technical SEO audit.

The FAQPage and HowTo deprecation

This is the part of a 2026 schema markup guide that most older articles still get wrong. Google restricted FAQ and HowTo rich results to a narrow set of authoritative government and health sites back in August 2023. That was a preview, not the end state. Google's own FAQPage documentation now carries a deprecation notice added on May 8, 2025: "This feature will no longer appear in Google Search starting May 7, 2026." The documentation itself was removed on June 15, 2025. If your site still ships a FAQPage block expecting the old expandable snippet in Google Search, that expectation is wrong now, for any property, not just the ones outside government and health.

Schema for AI and answer engines, a different game

None of this makes FAQ schema worthless. Its job changed. AI answer engines read the same JSON-LD block Google Search used to reward, and a clean, labeled question and answer pair is exactly the shape they quote from. Ahrefs' current guide makes a similar point about Organization, Author and Website schema: those types matter less for a classic rich result and more for establishing a clear brand entity that an AI system can attribute correctly, the same logic behind answer engine optimization generally. One implementation detail carries real weight here: schema injected by client-side JavaScript often never reaches an AI crawler like GPTBot, so the JSON-LD has to sit in the static HTML response, not get added after the page loads. For more on structuring content so an AI system quotes it directly, see how to get cited by ChatGPT, Perplexity and AI Overviews.

Tradeoffs and edge cases

Correct markup is necessary, never sufficient. Google states plainly that it doesn't guarantee a marked-up page will show a rich result, even when every required property is present and valid.

Structured data carries content policy risk too, not just a technical one. It has to describe content that's actually visible on the page. Marking up hidden or absent content, or using schema to mislead, is treated as a violation, not a gray area.

The most common practical mistake is a mismatched sameAs property, which can cause Google to misattribute a page to the wrong organization entirely. A related, dated trap: Google removed a large number of Knowledge Graph entries in early 2025, so a kgmid value that resolved correctly a year ago may not resolve now. Fix both the same way any audit would: verify the linked identifiers still point where they claim to, on a schedule, not just at launch.

Before publishing new markup, run it through Google's Rich Results Test to see what, if anything, it currently qualifies for, then through a generic validator like the Schema Markup Validator to confirm the JSON-LD itself is well formed. The two checks answer different questions. One confirms the syntax is valid schema.org markup. The other confirms Google will actually treat it as eligible for a rich result today. A block can pass the first check and fail the second, which is exactly what happened to every site still carrying FAQPage markup for the SERP feature it used to power.

Prioritization beats coverage. An e-commerce catalog gets the most out of Product, Breadcrumb and Organization schema. A blog or news property gets the most out of Article, Breadcrumb and Organization. A recipe site or a job board should go straight to the type built for it, Recipe or Job posting, rather than spreading effort across the full gallery. This is the same triage we run before any seo-aeo engagement at Kallos Labs: match the schema to what the page actually is, then verify it against the current gallery, because the list of what earns a rich result keeps moving.

Frequently asked questions

What is schema markup?

Schema markup is structured data, usually written as JSON-LD, that labels page content in a standardized vocabulary so search engines and AI systems can read it more precisely than plain HTML. It doesn't change what a visitor sees, it changes what a machine reader understands.

Which schema types still earn Google rich results in 2026?

Google's current gallery covers roughly 25 types, including Article, Product, Review snippet, Event, Job posting, Local business, Recipe, Video, Breadcrumb, Organization and Software app. FAQPage and HowTo are no longer on that list.

Does FAQ schema still do anything if Google dropped the rich result?

It no longer produces the expandable FAQ snippet in Google Search, but the same JSON-LD still gives AI answer engines a clean, labeled question and answer block to quote from, a separate benefit from the classic search rich result it used to power.

Does adding correct schema guarantee a rich result?

No. Google states directly that marking up a page correctly doesn't guarantee the rich result will display. Eligibility is necessary, never sufficient.