---
title: "FAQ schema for AI answers: writing FAQPage JSON-LD that parses"
slug: "faq-schema-for-ai-answers"
category: "structured-data"
canonical_path: "/articles/structured-data/faq-schema-for-ai-answers"
meta_title: "FAQ schema for AI answers, explained — Prime AI Visibility"
meta_description: "How to write FAQ content and FAQPage JSON-LD that parses cleanly — the 2023 Google rich-result restriction, question format, and copy-paste examples."
author: "The Prime AI Visibility editorial team"
date: "2026-07-31"
last_updated: "2026-07-31"
read_time: "10 min"
keywords:
  - FAQ schema
  - FAQPage JSON-LD
  - AI answers
  - question format
  - featured snippets
featured_image: "/brand/articles/structured-data/faq-schema-for-ai-answers.png"
featured_image_alt: "Nested open brackets unfolding into a fan of paired concentric arcs, one pair linked by a citrine node to a floating hexagon"
og_image: "/brand/articles/structured-data/faq-schema-for-ai-answers.og.png"
cta_mid_headline: "Which engines mention your FAQ?"
cta_mid_body: "Prime AI Visibility tracks whether answer engines reference your pages — and who they reference instead — across ChatGPT, Perplexity, Gemini, Claude, Copilot, and Google AI Overviews, checked every 24 hours."
cta_mid_button: "Check your visibility"
cta_bottom_headline: "Ship the markup, then measure what happens."
cta_bottom_body: "FAQPage JSON-LD is a means, not the metric. Create a workspace, add ten buyer prompts, and see how the engines respond over time."
cta_bottom_button: "Start tracking free"
---

# FAQ schema for AI answers: writing FAQPage JSON-LD that parses

FAQ schema is FAQPage JSON-LD that wraps a list of question-and-answer pairs so machines can parse them unambiguously. Since Google's August 2023 restriction, it rarely produces a search rich result — but the markup remains valid and makes clean question-and-answer text easier for any parser to isolate. Write real answers first, mark them up second, and the structure stays legible.

## FAQ schema for AI answers: the short answer

1. **The content is the asset; the markup is the label.** A well-structured FAQ with self-contained answers is the thing worth reading — FAQPage JSON-LD just makes the boundaries between question and answer machine-obvious.
2. **Google's 2023 change did not kill FAQ schema.** On August 8, 2023 Google limited FAQ rich results to well-known government and health sites, but the structured data is still valid and still validated by Google's own tooling.
3. **Question format matters.** Phrase each question the way a buyer types it, and answer it completely in the first two sentences — a long-standing editorial best practice for clarity, and the shape featured snippets have historically favored.

## What FAQ schema is (and what changed in 2023)

FAQ schema is a specific application of [schema.org](https://schema.org/FAQPage): a page typed as `FAQPage` containing an array of `Question` items, each with a nested `acceptedAnswer` of type `Answer`. It describes a page where a single author publishes both the questions and their answers — a product FAQ, a documentation page, a support article. That is distinct from `QAPage`, which describes a community forum thread where users post a question and other users compete to answer it. If your page is one voice answering its own questions, `FAQPage` is correct; if it is a Stack Overflow-style thread, `QAPage` is correct.

For years, marking up an FAQ earned a coveted expandable rich result directly in Google's search results. That era ended on August 8, 2023, when Google announced it would show FAQ rich results only for "well-known, authoritative government and health websites." For everyone else, the rich result disappeared from the standard search listing. Many teams read that announcement as "FAQ schema is dead" and stripped the markup out. That was the wrong lesson.

The markup did not become invalid — Google still validates FAQPage in its structured-data tooling, the schema.org type still exists, and crawlers can still parse it. What changed is only the *visual treatment* in one surface: the classic ten-blue-links search page. Meanwhile, a newer set of surfaces — ChatGPT, Perplexity, Gemini, Claude, Copilot, and Google's own AI Overviews — synthesize answers from retrieved content. These engines do not document exactly how (or whether) they weigh FAQPage markup, but clean, clearly labeled question-and-answer pairs are plausibly easier for any retrieval system to isolate than the same text buried in page chrome.

## Why FAQ markup may help retrieval

Retrieval-augmented systems generally do not read a page the way a browser renders it. The common public description of how such systems work — from vendor research and standard information-retrieval literature — is that they fetch content, split it into passages, embed those passages, and retrieve the ones most relevant to a prompt before synthesizing an answer. The specific engines named above do not publish the internals of their pipelines, so treat the mechanism below as a plausible model rather than documented per-engine behavior. On that model, two properties of an FAQ plausibly help:

- **Self-contained passages.** A good FAQ answer stands on its own without the surrounding page. If a passage is scored in isolation, as retrieval literature describes, an answer that depends on three paragraphs above it is weaker than one that reads completely on its own.
- **Explicit question-to-answer mapping.** FAQPage JSON-LD states, in machine terms, "this string is a question and this string is its answer." A parser reading the structured data does not have to infer that boundary from visual layout, which HTML scraping frequently gets wrong. Whether a given engine reads the JSON-LD or only the visible text is not documented.

This is the same discipline that governs [writing content ChatGPT will quote](https://primeaivisibility.com/articles/geo/how-to-write-content-chatgpt-will-quote): lead with the answer, keep it self-contained, and make the structure unmistakable. FAQ schema is that discipline expressed in a format machines can parse without guessing. For the broader picture of how these systems are understood to read markup, start with the pillar on [how AI answer engines consume structured data](https://primeaivisibility.com/articles/structured-data/structured-data-for-ai-search).

It is worth being honest about causation. There is no public evidence that FAQPage markup on its own produces any citation, ranking, or visibility gain, and no major engine documents such an effect. Do not trust anyone who promises a fixed lift. The plausible mechanism is indirect: the markup makes your best answers easier to parse, and the writing discipline it forces makes those answers clearer. Whether any of that changes how engines treat your page is an empirical question you can only answer by measuring over time.

## Write the answer first, then the markup

The single most common mistake is writing the JSON-LD before writing good FAQ content. Reverse it. The prose is the product; the schema is packaging.

A clear FAQ answer follows a tight pattern:

1. **State the answer in the first sentence.** No throat-clearing, no "it depends," no restating the question. If the reader stops after one sentence, they should have the answer.
2. **Add one or two sentences of qualification or proof.** The nuance, the exception, the number, the source. This is where "it depends" lives — after the direct answer, never before it.
3. **Keep it self-contained.** Do not reference "the table above" or "as we discussed." Write each answer so it reads completely on its own.
4. **Cap it around 40 to 60 words.** Long enough to be complete, short enough to read at a glance. This length range is an editorial best practice for readability and concision — a house guideline, not a documented threshold that any engine or featured-snippet system publishes.

Question format matters just as much. Phrase each question the way a real buyer types or speaks it — "How much does FAQ schema cost to implement?" beats "Implementation costs." Matching the natural-language query is a durable copywriting principle: it makes the question findable by humans and, plausibly, easier to match to a prompt, though engines do not document that matching. Comparison, definition, and cost questions tend to map directly to the queries people bring to answer engines. To see how surfaces can differ in what they surface from the same content, the breakdown of [how Perplexity and Google AI Overviews differ](https://primeaivisibility.com/articles/geo/perplexity-vs-google-ai-overviews) is a useful contrast.

## The FAQPage JSON-LD, minimally

Once the content is right, the markup is mechanical. A conforming FAQPage looks like this:

```json
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "Does FAQ schema still work after Google's 2023 change?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Yes, as valid structured data. Only the search rich result was restricted to government and health sites in August 2023; the markup itself remains valid."
      }
    },
    {
      "@type": "Question",
      "name": "What is the difference between FAQPage and QAPage?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "FAQPage is one author answering their own questions. QAPage is a community thread where users post questions and other users answer them."
      }
    }
  ]
}
```

A few rules keep this valid and honest:

- **Every visible question must have a visible answer on the page.** Google's guidelines require that the FAQ content marked up in JSON-LD also appears to the user. Marking up content that is hidden or absent risks a manual action and, more practically, produces markup that describes a page that does not exist.
- **`acceptedAnswer.text` may contain limited HTML** (links, lists, bold), but keep it simple. Plain, self-contained sentences are easier for any parser to handle than richly nested markup.
- **One FAQPage per page.** Do not scatter multiple FAQPage blocks across a single URL; consolidate the questions into one `mainEntity` array.
- **Do not mark up promotional or advertising text as an answer.** Google's policy explicitly disallows it, and it degrades the passage anyway.

JSON-LD in a `<script type="application/ld+json">` block is the format Google explicitly recommends in its structured-data documentation, because it keeps the structured data decoupled from the visible DOM. The consumer AI answer engines do not publish a recommended format at all. If you are weighing formats, the comparison of [JSON-LD versus microdata for machine consumption](https://primeaivisibility.com/articles/structured-data/json-ld-vs-microdata-for-ai) explains why inline attribute markup is generally harder to parse cleanly.

## Common misconceptions

**"Google killed FAQ schema, so remove it."** Google restricted the *rich result*, not the *schema*. The structured data is still valid, still validated by Google's tooling, and can still be parsed by any crawler. Removing it discards machine-readable structure in response to a cosmetic change in one surface.

**"More questions means more coverage."** Padding an FAQ with twenty thin questions dilutes the strong ones and makes the page worse to read. Six genuinely-asked questions with complete answers serve a reader better than twenty near-duplicates. Every question should map to a real query a buyer brings.

**"The markup is what matters."** The *answer text* is what a reader — or a model summarizing the page — actually uses. The markup only makes that text easier to locate and delimit. Perfect JSON-LD around a vague answer adds nothing of value.

**"FAQ schema guarantees a featured snippet."** It does not, and Google does not document schema as a snippet-selection factor. Featured snippets are selected algorithmically; there is no public evidence that FAQPage markup causes selection. Clear, well-structured answers are worth writing on their own merits.

## A pre-publish checklist

Before you ship an FAQ page, walk this list:

- Each question is phrased as a real buyer would type it, not as a heading fragment.
- Each answer states the answer in sentence one and is complete within roughly 40 to 60 words (a readability guideline, not an engine requirement).
- Every answer is self-contained — no "see above," no dependence on surrounding paragraphs.
- The visible page contains exactly the questions and answers in the JSON-LD, verbatim.
- The page uses `FAQPage` (self-authored) rather than `QAPage` (community thread), or the reverse if it is a forum.
- The JSON-LD validates in Google's Rich Results Test and Schema.org validator with no errors.
- There is one FAQPage block per URL, not several.
- No promotional copy is dressed up as an answer.

## FAQPage versus QAPage versus plain prose

Ratings below describe how *machine-parseable* the structure is — how unambiguous the question-to-answer boundary is for any parser. They do not describe, or predict, how a specific engine treats the page; no engine documents that.

| Approach | Best for | Parse clarity | Featured-snippet history |
|---|---|---|---|
| **FAQPage JSON-LD** | Self-authored product/support FAQs | Strong — explicit Q-to-A mapping | Strong before the 2023 restriction |
| **QAPage JSON-LD** | Community threads, forums | Partial — many answers, one question | Partial |
| **Plain Q&A prose (no schema)** | Any page with clear headings | Partial — parser must infer structure | Strong if headings are clean |

The table's point is that plain, well-structured prose is already readable and parseable — schema clarifies structure, it is not a prerequisite. If you only ever ship clean `## question` headings followed by direct-answer paragraphs, the structure is still legible. FAQ schema simply removes ambiguity about where a question ends and its answer begins. To keep the vocabulary of extraction, retrieval, and citation straight as you build, the [structured-data and GEO glossary](https://primeaivisibility.com/glossary) is the fastest reference; and the [teams-and-workflows view of citation tracking](https://primeaivisibility.com/use-cases) shows how support, content, and SEO teams actually use this data day to day.

<!-- cta:mid -->

> **Which engines mention your FAQ?**
>
> Prime AI Visibility tracks whether answer engines reference your pages — and who they reference instead — across ChatGPT, Perplexity, Gemini, Claude, Copilot, and Google AI Overviews, checked every 24 hours.
>
> **[Check your visibility](https://app.primeaivisibility.com/sign-up)**

<!-- /cta:mid -->

## References

1. Google Search Central, *Changes to HowTo and FAQ rich results* (August 8, 2023). <https://developers.google.com/search/blog/2023/08/howto-faq-changes>
2. Google Search Central, *FAQ (FAQPage) structured data documentation*. <https://developers.google.com/search/docs/appearance/structured-data/faqpage>
3. Google Search Central, *Q&A (QAPage) structured data documentation*. <https://developers.google.com/search/docs/appearance/structured-data/qapage>
4. Schema.org, *FAQPage type definition*. <https://schema.org/FAQPage>
5. Schema.org, *Question type definition*. <https://schema.org/Question>

## Next steps

1. **[See how the pillar explains AI structured-data consumption](https://primeaivisibility.com/articles/structured-data/structured-data-for-ai-search)** so your FAQ markup fits a coherent schema strategy rather than sitting alone.
2. **[Learn the writing discipline behind quotable content](https://primeaivisibility.com/articles/geo/how-to-write-content-chatgpt-will-quote)** before you write a single answer — clarity is the part you control.
3. When your FAQ is live, **[create a Prime AI Visibility workspace](https://app.primeaivisibility.com/sign-up)** and bring 10 buyer prompts to measure how the engines respond over time.

## Frequently asked questions

**Does FAQ schema still work after Google's 2023 change?**
Yes, as valid structured data. On August 8, 2023 Google restricted the FAQ *rich result* — the expandable snippet in search results — to well-known government and health sites. The FAQPage structured data itself remains valid and is still validated by Google's tooling. Whether the consumer AI answer engines use it is not something they document.

**What is the difference between FAQPage and QAPage schema?**
FAQPage describes a page where one author publishes both the questions and their answers, like a product or support FAQ. QAPage describes a community thread where a user posts a question and other users post competing answers, like a forum. Use FAQPage for self-authored FAQs and QAPage for forum-style content.

**Do AI answer engines read FAQ schema?**
It is not publicly documented. The major consumer engines — ChatGPT, Perplexity, Gemini, Claude, Copilot, and Google AI Overviews — do not publish whether or how they weigh FAQPage markup. What the markup verifiably does is give any parser an explicit mapping from each question to its answer, which removes ambiguity. There is no public evidence of a guaranteed visibility or citation gain from the markup alone, so treat it as a parse aid, not a ranking lever.

**How should I phrase FAQ questions for featured snippets and AI answers?**
Phrase each question the way a real buyer types or speaks it, matching natural-language queries — "How much does X cost?" rather than "Pricing." Then answer completely in the first one or two sentences, keep the answer self-contained, and cap it around 40 to 60 words. This is an editorial best practice for clarity and matching real queries; engines do not document it as a requirement, and it does not guarantee a snippet or a mention.

**Can I mark up FAQ content that is hidden behind an accordion?**
Yes, as long as the content is present on the page and available to the user when they interact with it. Google's guidelines require that the questions and answers in your JSON-LD also appear in the visible page. Marking up content that is entirely absent or that users can never see risks a manual action.

**Is FAQPage JSON-LD better than microdata for AI consumption?**
JSON-LD is the format Google recommends in its structured-data documentation because it keeps the data in a single, decoupled script block rather than scattering it across DOM attributes as microdata does. Decoupled blocks are generally easier to parse, and the consumer AI engines do not publish a preference either way. For a full comparison, see the dedicated article on JSON-LD versus microdata for AI.

<!-- cta:bottom -->

> **Ship the markup, then measure what happens.**
>
> FAQPage JSON-LD is a means, not the metric. Create a workspace, add ten buyer prompts, and see how the engines respond over time.
>
> **[Start tracking free](https://app.primeaivisibility.com/sign-up)**

<!-- /cta:bottom -->


<!-- structured-data -->
<script type="application/ld+json">{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://primeaivisibility.com/#organization","name":"Prime AI Visibility","url":"https://primeaivisibility.com/","mainEntityOfPage":{"@id":"https://primeaivisibility.com/about#webpage"},"logo":"https://primeaivisibility.com/brand/logos/citorum-wordmark-ink-on-cream@2x.png","description":"Prime AI Visibility tracks how often your brand is cited, recommended, and quoted across every major AI answer engine.","slogan":"Be the answer, not the runner-up.","foundingDate":"2025","email":"hello@primeaivisibility.com","sameAs":["https://app.primeaivisibility.com/"],"contactPoint":[{"@type":"ContactPoint","contactType":"customer support","email":"hello@primeaivisibility.com","url":"https://primeaivisibility.com/about","availableLanguage":["English"]},{"@type":"ContactPoint","contactType":"press","email":"press@primeaivisibility.com","url":"https://primeaivisibility.com/about"},{"@type":"ContactPoint","contactType":"privacy","email":"privacy@primeaivisibility.com","url":"https://primeaivisibility.com/privacy"}]},{"@type":"Person","@id":"https://primeaivisibility.com/about#editorial-team","name":"The Prime AI Visibility editorial team","url":"https://primeaivisibility.com/about","jobTitle":"Editorial team","worksFor":{"@id":"https://primeaivisibility.com/#organization"},"knowsAbout":["Generative Engine Optimization","Share of citation","Retrieval-augmented generation","AI answer engines"]},{"@type":"WebSite","@id":"https://primeaivisibility.com/#website","url":"https://primeaivisibility.com/","name":"Prime AI Visibility","publisher":{"@id":"https://primeaivisibility.com/#organization"},"inLanguage":"en-US"},{"@type":"SoftwareApplication","@id":"https://primeaivisibility.com/#software","name":"Prime AI Visibility","applicationCategory":"BusinessApplication","operatingSystem":"Web","url":"https://primeaivisibility.com/","description":"Generative Engine Optimization (GEO) platform that monitors brand citations across ChatGPT, Perplexity, Gemini, Claude, Copilot, and Google AI Overviews.","publisher":{"@id":"https://primeaivisibility.com/#organization"},"offers":{"@type":"Offer","url":"https://app.primeaivisibility.com/sign-up","category":"SaaS subscription"}}]}</script>
<script type="application/ld+json">{"@type":"BlogPosting","@id":"https://primeaivisibility.com/articles/structured-data/faq-schema-for-ai-answers#article","mainEntityOfPage":"https://primeaivisibility.com/articles/structured-data/faq-schema-for-ai-answers","headline":"FAQ schema for AI answers: writing FAQPage JSON-LD that parses","description":"How to write FAQ content and FAQPage JSON-LD that parses cleanly — the 2023 Google rich-result restriction, question format, and copy-paste examples.","datePublished":"2026-07-31","dateModified":"2026-07-31","inLanguage":"en-US","image":"https://primeaivisibility.com/brand/articles/structured-data/faq-schema-for-ai-answers.og.png","author":{"@type":"Organization","@id":"https://primeaivisibility.com/#organization","name":"Prime AI Visibility","url":"https://primeaivisibility.com/about"},"publisher":{"@id":"https://primeaivisibility.com/#organization"},"keywords":["FAQ schema","FAQPage JSON-LD","AI answers","question format","featured snippets"],"articleSection":"structured-data"}</script>
<script type="application/ld+json">{"@type":"BreadcrumbList","@id":"https://primeaivisibility.com/articles/structured-data/faq-schema-for-ai-answers#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://primeaivisibility.com/"},{"@type":"ListItem","position":2,"name":"Journal","item":"https://primeaivisibility.com/articles"},{"@type":"ListItem","position":3,"name":"FAQ schema for AI answers: writing FAQPage JSON-LD that parses","item":"https://primeaivisibility.com/articles/structured-data/faq-schema-for-ai-answers"}]}</script>
<script type="application/ld+json">{"@type":"FAQPage","@id":"https://primeaivisibility.com/articles/structured-data/faq-schema-for-ai-answers#faq","mainEntity":[{"@type":"Question","name":"Does FAQ schema still work after Google's 2023 change?","acceptedAnswer":{"@type":"Answer","text":"Yes, as valid structured data. On August 8, 2023 Google restricted the FAQ *rich result* — the expandable snippet in search results — to well-known government and health sites. The FAQPage structured data itself remains valid and is still validated by Google's tooling. Whether the consumer AI answer engines use it is not something they document."}},{"@type":"Question","name":"What is the difference between FAQPage and QAPage schema?","acceptedAnswer":{"@type":"Answer","text":"FAQPage describes a page where one author publishes both the questions and their answers, like a product or support FAQ. QAPage describes a community thread where a user posts a question and other users post competing answers, like a forum. Use FAQPage for self-authored FAQs and QAPage for forum-style content."}},{"@type":"Question","name":"Do AI answer engines read FAQ schema?","acceptedAnswer":{"@type":"Answer","text":"It is not publicly documented. The major consumer engines — ChatGPT, Perplexity, Gemini, Claude, Copilot, and Google AI Overviews — do not publish whether or how they weigh FAQPage markup. What the markup verifiably does is give any parser an explicit mapping from each question to its answer, which removes ambiguity. There is no public evidence of a guaranteed visibility or citation gain from the markup alone, so treat it as a parse aid, not a ranking lever."}},{"@type":"Question","name":"How should I phrase FAQ questions for featured snippets and AI answers?","acceptedAnswer":{"@type":"Answer","text":"Phrase each question the way a real buyer types or speaks it, matching natural-language queries — \"How much does X cost?\" rather than \"Pricing.\" Then answer completely in the first one or two sentences, keep the answer self-contained, and cap it around 40 to 60 words. This is an editorial best practice for clarity and matching real queries; engines do not document it as a requirement, and it does not guarantee a snippet or a mention."}},{"@type":"Question","name":"Can I mark up FAQ content that is hidden behind an accordion?","acceptedAnswer":{"@type":"Answer","text":"Yes, as long as the content is present on the page and available to the user when they interact with it. Google's guidelines require that the questions and answers in your JSON-LD also appear in the visible page. Marking up content that is entirely absent or that users can never see risks a manual action."}},{"@type":"Question","name":"Is FAQPage JSON-LD better than microdata for AI consumption?","acceptedAnswer":{"@type":"Answer","text":"JSON-LD is the format Google recommends in its structured-data documentation because it keeps the data in a single, decoupled script block rather than scattering it across DOM attributes as microdata does. Decoupled blocks are generally easier to parse, and the consumer AI engines do not publish a preference either way. For a full comparison, see the dedicated article on JSON-LD versus microdata for AI."}}]}</script>
<!-- /structured-data -->