FAQ Schema: How to Structure Q&As for Google, AI Overviews & LLM Answers

What is FAQ schema, exactly?

FAQ schema is structured data — code added to a page's HTML — that tells search engines and AI systems "this block of content is a list of questions, each followed by its answer." It's built on Schema.org's FAQPage type, the shared vocabulary search engines use to parse what a page actually contains, rather than guessing from raw text.

Done well, it turns a plain list of Q&As into something machine-readable: a crawler — or an AI model retrieving live content — can lift a single question and its exact answer without reading the whole page. That's the whole value proposition. You're not changing what a visitor sees; you're giving every system that reads your page a cleaner way to extract the parts that answer a specific query.

Where FAQ schema shows up:

• People Also Ask
• AI Overviews
• ChatGPT & Perplexity
• Voice assistants

The 2023 rule change most guides still ignore

Most FAQ schema articles still show screenshots of expandable FAQ accordions sitting under every other blue link. That's outdated. In August 2023, Google restricted the visual FAQ rich result in classic search to a small set of authoritative government and health websites. For nearly everyone else, marking up a page no longer produces that collapsible snippet in the regular results.

So should you bother with FAQ schema in 2026? Yes — just for a different reason. The schema is no longer primarily a play for extra SERP real estate. It's content infrastructure: the clearest, most explicit signal you can give any system — Google, Bing, an AI crawler, a voice assistant — about which text answers which question. The rich snippet was one output of that structure. It was never the only one, and for AI answer engines it was never even the main one.

FAQPage vs. QAPage: pick the right markup

This is where most implementations go wrong before they even start. The two schema types look similar but describe different page formats, and Google's guidelines are explicit that mixing them up is invalid usage.

Criteria	FAQPage	QAPage
Best for	A page with a fixed list of questions the site itself answers	A page with one question and multiple community-submitted answers
Who answers	The site/author, one accepted answer per question	Any user; several candidate answers, one often marked "best"
Can users add alternate answers?	No	Yes — that's the defining feature
Typical page	FAQ hub, product support page, knowledge-base article	Forum thread, Q&A community page
Wrong use case	A forum page with multiple user answers	A single-answer FAQ page written by the site

If you're writing the FAQ yourself — which covers almost every business blog, support page and landing page — FAQPage is the correct choice. QAPage exists for forums and community Q&A platforms where visitors post competing answers.

Google's content guidelines, condensed

Google will ignore or penalize markup that doesn't meet its structured-data policies.

Valid use:

• A support or knowledge-base page listing real, site-authored questions and answers
• The full question text and the full answer text, both included in the markup
• Content that's visible to every visitor on the live page, not hidden behind a login or script

Invalid use:

• Forum or product pages where users submit their own answers (use QAPage instead)
• FAQ markup used purely for advertising or promotional copy
• Obscene, hateful, violent or illegal content inside any question or answer
• The same FAQ block duplicated across dozens of unrelated pages

How to add FAQ schema in JSON-LD

JSON-LD is what Google recommends: it's a single script block you drop into the page, decoupled from your HTML, so a design change can't accidentally break it.

1. Pull real questions, not invented ones. Use your own search queries, support tickets, sales calls and the "People also ask" box for your target keyword. Three to five genuine questions outperform ten generic ones.
2. Write direct, two-to-three sentence answers. Lead with the actual answer in the first sentence — don't bury it under context. This is the sentence most likely to get lifted into an AI Overview or voice response.
3. Generate the JSON-LD. Hand-write it for a handful of questions, or use a free generator for speed. Keep question and answer text identical to what's visible on the page.
4. Add the script to the page. Place the <script type="application/ld+json"> block in the <head> or just before </body>. One FAQPage block per page, maximum.
5. Validate, then request indexing. Test it, fix any errors, then ask Google Search Console to recrawl the page.

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is FAQ schema?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "FAQ schema is structured data that marks up a list of questions and answers so search engines and AI systems can read and cite them directly."
      }
    },
    {
      "@type": "Question",
      "name": "Does FAQ schema still work in 2026?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Yes, though the visual rich snippet is now limited to a small set of sites. The structured data still helps AI Overviews, LLMs and voice assistants parse and cite your content."
      }
    }
  ]
}

Prefer microdata?

Microdata embeds the same information as HTML attributes instead of a separate script. It works, but Google recommends JSON-LD, and mixing both formats on one page is asking for conflicting signals — pick one and stay consistent sitewide.

<div itemscope itemtype="https://schema.org/FAQPage">
  <div itemscope itemprop="mainEntity" itemtype="https://schema.org/Question">
    <h3 itemprop="name">What is FAQ schema?</h3>
    <div itemscope itemprop="acceptedAnswer" itemtype="https://schema.org/Answer">
      <p itemprop="text">Structured data marking up Q&A content.</p>
    </div>
  </div>
</div>

Testing and validating your markup

Google retired the old Structured Data Testing Tool — if a guide still points you there, it's out of date. Use these two instead:

• Rich Results Test (search.google.com/test/rich-results) — checks whether a URL or code snippet is eligible for Google's rich results and flags markup errors.
• Schema Markup Validator (validator.schema.org) — checks general Schema.org compliance, useful for markup Google doesn't have a dedicated rich-result test for.

Once it's clean, open Google Search Console, paste the page URL into the top search bar, and click Request Indexing so Google recrawls the page with the new markup in place.

Common mistakes that get FAQ schema rejected

• Curly quotes from Word or Docs. Pasting text straight from a word processor often swaps straight quotes for "smart" ones, which silently breaks JSON syntax.
• Markup that doesn't match the visible page. If the schema text and the on-page text diverge, Google treats it as misleading markup.
• The same FAQ block pasted on every page. Reserve it for pages where those specific questions genuinely belong.
• Promotional copy disguised as an answer. "Why is our product the best?" with a sales pitch for an answer violates Google's guidelines directly.
• More than one FAQPage block per page. Consolidate every question into a single block.

PAA, AI Overviews, voice search and LLMs

People Also Ask

Schema doesn't guarantee a PAA placement — Google generates that box from its own analysis of the whole web, not from any one page's markup. But the discipline FAQ schema forces (one clear question, one direct answer) is exactly the format PAA rewards, so sites that mark up real questions tend to show up there more often.

AI Overviews and LLM answer engines

This is where FAQ schema earns its keep in 2026. Generative answer engines retrieve and summarize web content on the fly, and they favor sources that are unambiguous about what answers what. A clean Question → Answer pair is easier to extract and cite correctly than the same information buried in a paragraph — which is why structured, well-marked content is a core piece of LLM optimization broadly, not just classic SEO.

Voice search

Voice assistants mostly answer from featured snippets and short, extractable text. The same direct-answer-first writing that helps FAQ schema succeed with AI Overviews helps it surface in voice results too — there's no separate format to design for.

Implementation checklist

• Confirmed the page is genuinely a site-authored FAQ, not a forum (use FAQPage, not QAPage)
• 3–5 real questions sourced from search data, support tickets or sales calls
• Answers lead with the direct answer in the first sentence
• Question and answer text in the schema exactly matches the visible page
• JSON-LD written in straight quotes, one FAQPage block on the page
• Passed the Rich Results Test and the Schema Markup Validator with no errors
• Requested indexing in Google Search Console for the updated page