FAQPage Schema for AI Search: Implementation + 7 Cited Examples

Does FAQPage schema still work after Google deprecated FAQ rich results?

Yes. Google deprecated visible FAQ rich results, not the FAQPage schema type itself. On August 8, 2023, Google announced FAQ rich snippets would only appear for "well-known, authoritative government and health websites." The schema continues to be valid, indexed, and parsed by AI engines.

The distinction matters. SEO blogs treated the deprecation as a death notice. AI engines did not.

• ChatGPT tokenizes FAQPage JSON-LD as raw text inside the page, contributing to retrieval-augmented generation context.
• Perplexity weights structured Q&A pairs heavily because its citation engine maps user prompts to extractable answer blocks.
• Google AI Overviews uses structured data as a ranking signal for which passages get surfaced, even though the visible blue rich result is gone.

Google's own guidance from the deprecation post: "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." Translation: keep it.

FAQPage shifted from an SEO tactic (CTR boost via SERP real estate) to an AEO tactic (extraction signal for answer engines). The schema is the same. The payoff moved.

What's the correct JSON-LD format for FAQPage schema in 2026?

The correct FAQPage JSON-LD uses @context: https://schema.org, @type: FAQPage, and a mainEntity array of Question objects, each containing a name (the question text) and an acceptedAnswer of @type: Answer with a text property. One FAQPage block per page. Place it in a <script type="application/ld+json"> tag in the page head.

Here is the canonical pattern, current with the Schema.org FAQPage spec:

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "Does FAQPage schema still work for AI search?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Yes. Google deprecated visible FAQ rich results in August 2023, but FAQPage JSON-LD remains a valid extraction signal for ChatGPT, Perplexity, and Google AI Overviews. Pages with FAQPage markup are 3.2x more likely to appear in AI Overviews."
      }
    },
    {
      "@type": "Question",
      "name": "How many FAQs should a page have?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "5 to 10 visible question-answer pairs is the sweet spot. Fewer than 5 limits extraction surface area. More than 10 dilutes topical focus and risks looking like keyword stuffing to spam classifiers."
      }
    }
  ]
}

Mandatory rules from the Google FAQ structured data docs:

• Both name and acceptedAnswer.text are required.
• HTML inside text is allowed (<p>, <a>, <ul>, <li>, <br>), but escape it properly.
• Every Q&A in the schema must appear verbatim or near-verbatim in the visible HTML.
• One FAQPage per URL. Multiple blocks confuse parsers and trigger validation errors.

Validate with the Schema.org Validator before publishing. Google's Rich Results Test will report "page is not eligible for rich results," which is expected post-2023 and not an error.

Where on a page should FAQs live for maximum AI citation?

Place the FAQ block near the bottom of the article body, before related content and after the main answer sections. AI engines weight FAQ content most heavily when it appears as a structured summary of the page's key questions, not as the page's only content.

Three placement rules from auditing cited B2B pages:

1. After the main body, before the footer. Pages cited by AI Overviews almost universally place FAQs in the final 25% of the article. The model has already extracted context from the body and uses the FAQ block to confirm or refine specific answers.
2. Visible by default, no accordion-only patterns. If your FAQ uses an accordion that loads answers via JavaScript only on click, server-side rendered HTML may not contain the answer text. Crawlers see empty <div>s. Render answers in the initial HTML, then progressively enhance with collapse behavior.
3. One FAQ block per URL, deduplicated across the site. Repeating the same FAQ block on 50 pages is a known spam pattern. Make each page's FAQs page-specific.

The TL;DR / answer-first rule still applies: do not bury the page's primary answer behind the FAQ. Lead with the answer in the first 100 words (90% of top-cited pages do this), then use the FAQ block to capture long-tail and adjacent queries.

How many FAQs should a page have? Is there a sweet spot?

5 to 10 question-answer pairs per page is the optimal range, with each answer scoped to 40-60 words. This balances three competing pressures: enough surface area for AI extraction, tight enough to maintain topical focus, and short enough that each answer reads as a confident, citation-ready unit.

The data:

• Averi's FAQ optimization guidance: "Include 5-10 FAQ questions per page for pillar content. Fewer than 5 provides limited value for AI extraction. More than 10 dilutes focus."
• Frase.io reports 78% of AI-generated answers use list formats, and FAQ blocks of 5-10 items match that extraction shape.
• Pages with 20+ FAQs frequently trip duplicate-content and keyword-stuffing classifiers, especially when answers exceed 200 words each.

Quality beats quantity. Three razor-sharp FAQs that exactly match high-intent queries outperform 20 generic ones. Pull questions from:

• People Also Ask boxes for your primary keyword.
• Reddit and Quora threads on the topic (Perplexity weights Reddit at 46.7% of citations).
• Your sales team's lost-deal objections. These are the highest-intent questions in your category and rarely covered by competitors.
• AI engine prompt variations. Ask ChatGPT and Perplexity the page's core question 5-10 different ways. Each rephrase is a candidate FAQ.

Can you add FAQPage schema to a non-FAQ page?

Yes, as long as the questions and answers are visible on the page, even if they are not in a section labeled "FAQ." Google explicitly confirmed this: "If the questions and answers are visible on the page, even if they're in different places on the page, that's perfectly fine."

This is the highest-leverage move most B2B sites miss. You can add FAQPage schema to:

• Pricing pages that include "How does billing work?", "What happens at renewal?", "Can I cancel anytime?".
• Product pages that answer "What integrations are supported?", "How long does setup take?".
• Long-form blog posts where H2s like "What is X?", "How does X work?", and "When should you use X?" already function as questions.

Three constraints:

1. Visibility is non-negotiable. Both the question and answer must render in the page HTML. Hidden, accordion-only, or modal-only content violates Google's structured data spam policy and can trigger a manual action.
2. Single accepted answer per question. FAQPage assumes a definitive answer. If your page has user-submitted multi-answer threads, use QAPage instead.
3. Do not retrofit advertising as FAQs. Promotional copy disguised as a question ("Why is [Brand] the best?") fails validation and reduces extraction quality.

What are the 5 most common FAQPage implementation errors?

Most FAQPage implementations fail one of five validation or extraction tests. Each error either invalidates the schema entirely or quietly suppresses AI citation likelihood.

1. Schema content does not match visible content. The JSON-LD declares 8 questions; the page only shows 4. Or the answer text in the schema is fuller than what users see. This is the most common spam-policy violation and risks a manual action that wipes rich-result eligibility for the entire domain.

2. Using FAQPage on community Q&A or forum content. FAQPage requires a single accepted answer authored by the site. Forums with user-submitted answers and voting must use QAPage instead. Mismatching the type breaks parsing.

3. Duplicating identical FAQ schema across many pages. Copy-pasting the same block across 50 pages signals manipulation. Each FAQPage block must be page-specific. If three pages legitimately share an FAQ, mark it on the canonical and link to it from the others.

4. Answers rendered only via JavaScript or behind closed accordions. Crawlers and AI extraction pipelines parse the initial server-rendered HTML. If your accordion loads answer text on click via client-side JS, the answer is invisible to the schema parser. Render the full text in HTML, then progressively enhance.

5. Improper nesting inside Article schema. Wrapping FAQPage inside Article.mainEntityOfPage creates invalid graph structure. The correct pattern is two sibling JSON-LD blocks (one Article, one FAQPage), or a single graph using @graph array notation.

Fix all five by running every published page through the Schema.org Validator and Google's Rich Results Test before going live.

How do you implement FAQPage schema step by step?

Implementing FAQPage schema for AI search takes 30-60 minutes per page. Follow these eight steps in order. Skipping the validation step is the single most common cause of silent failures.

1. Audit the page for question-shaped content. Identify 5-10 questions the page already answers, or that the page should answer based on People Also Ask, Reddit, and Perplexity prompts.
2. Write or rewrite each answer to 40-60 words. Lead with a direct answer, then add one supporting sentence. This is the extraction-friendly shape AI engines cite.
3. Add a visible FAQ section to the page HTML. Place it after the main body, before the footer. Use <h2> for "Frequently Asked Questions" and <h3> for each question.
4. Generate the JSON-LD block. Use the canonical pattern above or a generator like Saijo George's FAQ schema tool. One FAQPage per page.
5. Insert the JSON-LD into the page <head> inside <script type="application/ld+json"> tags. Place it after Article schema if both exist.
6. Verify the schema content matches the visible content verbatim. Run a manual diff. The AI parser will compare both.
7. Validate with Schema.org Validator and Google Rich Results Test. Resolve all errors. Warnings about "page not eligible for rich results" are expected post-2023 and are not failures.
8. Track AI citation in Profound, Otterly, or similar. Test the page's primary question against ChatGPT, Perplexity, and Google AI Overviews 7 days after publishing. Iterate on answer wording until cited.

For scale, add FAQPage generation to your CMS templates so every new article ships with valid schema by default.

Which 7 B2B sites get cited via FAQPage schema today?

These seven B2B FAQPage implementation patterns appear consistently in AI Overview and Perplexity citations as of 2026. Each represents a distinct strategy worth copying. View source on each page and search for "@type":"FAQPage".

#	Site / Pattern	What they do well	Worth copying
1	Stripe Docs -- developer FAQ blocks at the bottom of feature docs	Each feature page ends with 4-6 question-shaped headings answering integration edge cases. Schema renders server-side.	Embedded FAQs on documentation, not a separate FAQ page.
2	HubSpot Knowledge Base -- article-level FAQPage on every help doc	Every article carries Article + FAQPage schema. Answers scoped to 50-80 words.	Treats every help doc as a multi-question FAQ.
3	Notion Help Center -- FAQ blocks under feature explainers	Mixes how-to body with a 5-question FAQ summarizing the same content. Heavy citation rate in Perplexity.	Restating body content as Q&A doubles extractability.
4	Zapier blog -- listicle posts ending with FAQPage	Every "best X tools" listicle ends with a 6-question FAQPage covering pricing, alternatives, and use cases.	FAQPage on listicles, not just guides.
5	Atlassian Cloud pricing -- FAQPage on the pricing page	Pricing pages with billing, renewal, and seat-tier FAQs render schema for non-FAQ pages.	Pricing-page FAQs convert AND get cited.
6	Cloudflare Learning -- definitional FAQs on every concept page	Each "What is [networking concept]?" page is structured as one FAQPage with 8-12 nested questions. Dominates AI Overview citations for networking terms.	Encyclopedia-style FAQ pages outperform blog format.
7	Algolia documentation -- inline FAQ at end of every guide	Auto-generated FAQ block per doc page using their internal CMS template. Consistent schema, consistent citation.	Templated FAQPage at scale via CMS.

The pattern across all seven: FAQPage is not a standalone page type. It is a block embedded inside high-value pages (docs, pricing, listicles, learning resources) where the question-shaped content already exists.

What does the data say about FAQPage schema and AI citation rates?

Three credible sources triangulate the impact of FAQPage schema on AI search visibility. Numbers vary, but the direction is consistent: FAQPage materially lifts citation rates.

• 3.2x more likely to appear in Google AI Overviews. Frase.io research, citing Search Engine Land's 2024 analysis.
• 30% average lift in AI citation rates across major AEO platforms, with up to 35% on question-based queries. Stackmatix structured data benchmarks.
• 78% of AI-generated answers use list formats, the exact shape FAQPage produces. Frase, citing Princeton GEO research.
• Only 12.4% of websites implement structured data, per Stackmatix's 2026 schema audit. FAQPage adoption is even lower, which means the competitive bar to clear is unusually low.
• AI-referred sessions grew 527% between January and May 2025 (Search Engine Land), making the citation-rate uplift compound on a rapidly growing traffic base.

The combined math: a B2B page with valid FAQPage schema, in a category where less than 15% of competitors have implemented it, on a traffic source growing 5x year over year. That is the AEO equivalent of a backlink in 2010.

Pattern	Site Example	FAQ Placement	Schema Approach	Why It Gets Cited
Embedded doc FAQs	Stripe Docs	Bottom of feature docs	Article + FAQPage	Answers integration edge cases mapped to dev queries
KB article FAQs	HubSpot Knowledge Base	End of every help article	Article + FAQPage on every URL	Templated at scale, 50-80 word answers
Feature explainer FAQs	Notion Help Center	Under each feature explainer	FAQPage block summarizing body	Restates body as Q&A, doubles extractability
Listicle FAQs	Zapier blog	End of best-X-tools posts	Article + ItemList + FAQPage	Captures pricing, alternatives, use cases
Pricing-page FAQs	Atlassian pricing	Below pricing tiers	FAQPage on non-FAQ page	Billing/renewal questions cited in buyer queries
Learning-page FAQs	Cloudflare Learning	Throughout encyclopedia entries	Single FAQPage with 8-12 questions	Dominates definitional AI Overview citations
CMS-templated FAQs	Algolia documentation	Auto-generated per doc	Article + FAQPage via CMS template	Consistent schema enforcement at scale