TLDR: FAQ Schema Is One of the Highest-Correlated Signals for AI Citations
Google deprecated FAQ schema for search results because it stopped appearing in SERPs. Most SEO teams stopped implementing it. That was the right call for SEO. It was the wrong call for AEO.
Recent research on 10,000 URLs across 8 AI engines found FAQ schema correlates with a +24 lift on AI Overview citations. On-page FAQ sections add another +18. Combined, that is one of the highest-impact structural changes you can make. Every SaaS founder still ignoring FAQ schema is leaving citations on the table.
The rule: FAQ schema on every page. Different FAQs on every page. Copy-pasting the same 5 questions across your site is a tell for LLMs and dilutes the signal.
Why did Google deprecate FAQ schema in the first place?
Google announced in August 2023 that FAQ rich results would only display for authoritative government and health sites. Everyone else lost the visible SERP feature.
The interpretation across the SEO industry was: "FAQ schema is dead, stop implementing it." That interpretation was too broad. Google's change killed the SERP display, not the underlying schema. The JSON-LD still gets parsed by every crawler and every AI training pipeline. Ripping it out was an overreaction, and now the SEO teams that did are behind.
Reference: Google's FAQPage documentation confirms the schema itself is still valid. It just does not surface in Google's SERP anymore.
What does the data actually show about FAQ schema and AEO?
HubSpot's 2026 analysis of 10,000 URLs across 8 AI engines produced correlation data by signal. The FAQ-related findings:
- FAQ schema in JSON-LD: +24 lift on AI Overview citations
- FAQ sections on page: +18 lift on AI Overview citations
- Question-style headings (H2, H3): +28 on AI Overviews, +19 on Gemini
These three signals stack. A page with question-style H2s, an on-page FAQ section, and FAQ schema markup has cumulative lift that dominates other on-page factors. Meanwhile the biggest traditional SEO signal, backlinks, correlated at essentially zero across all 8 engines.
You can cross-reference with Search Engine Journal, Search Engine Land, Ahrefs blog, and Backlinko for ongoing coverage. The pattern is consistent across independent studies: structured Q&A wins for AI citations.
Why do AI engines love FAQ format so much?
Because AI engines are literally answer engines. The clue is in the name. Answer Engine Optimization. What is a FAQ? Frequently Asked Questions. Questions that lead to answers. That is exactly the format LLMs are trying to produce when they respond to a user query.
Every FAQ pair is a self-contained fact. Question at the top, answer directly below, no cross-references required. LLMs chunk content this way naturally. When ChatGPT is trying to answer "what is the difference between X and Y," it looks for extractable Q&A pairs that already answer that exact question. Your FAQ section is a pre-built answer library.
The ChatGPT Search documentation and Perplexity's blog both describe retrieval-augmented generation flows that reward this structure.
What is the FAQ-per-page rule and why does it matter?
Every page on your site should have a FAQ section. Different FAQs on different pages. That is the rule.
The biggest mistake teams make when they hear "FAQ schema on every page" is copy-pasting the same 5 questions across the entire site. The homepage has the same FAQ as the pricing page has the same FAQ as every blog post. Zero variation.
Why that fails: LLMs treat duplicate content as low signal. If your homepage FAQ and your blog post FAQ are identical, neither is seen as authoritative for the specific query that page is meant to answer. Both get discounted.
The right approach: each page's FAQ answers questions specific to that page's intent. Homepage FAQs cover product-level questions. Pricing page FAQs cover billing and plan questions. Blog post FAQs cover topic-specific questions the article should have answered but did not fully.
How do I write good FAQ questions?
Good FAQ questions match how users actually type into AI engines. Bad FAQ questions are made up by marketing to reinforce a message.
Good sources for real FAQ questions:
- Google Search Console: filter for 4 to 8 word queries with impressions but low CTR. Those are questions people are typing but not clicking your result.
- Answer The Public: pulls autocomplete data and clusters by question type (who, what, when, where, why, how).
- Customer support tickets: what actual customers ask before or after buying.
- Reddit and Quora threads in your niche: real user questions in real language.
- Sales call recordings: what prospects ask on discovery calls.
Bad sources:
- Marketing team brainstorms.
- The FAQ from your competitor (copy their positioning, do not copy their exact FAQs).
- Made-up questions that let you shoehorn in a keyword.
How do I implement FAQ schema in Laravel?
The Growth Pigeon article view already does this automatically. It parses the H2 FAQ block from the article body and emits FAQPage JSON-LD. Reference implementation:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "Your question here?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Your answer here."
}
}
]
}
</script>Reference: Schema.org FAQPage spec. Validate with Google Rich Results Test. If you are on WordPress, plugins like Yoast SEO or Rank Math add FAQ schema in their content blocks.
The technical implementation is 30 minutes. The content work (writing good FAQs per page) takes 30 to 60 minutes per page. That is where the real effort lives.
How many FAQ questions per page is optimal?
5 to 8 is the sweet spot for most articles and landing pages. Fewer than 3 and the schema does not carry enough weight. More than 10 and the individual questions dilute in weight and the page starts to feel padded.
For pillar articles (2,000+ words), 7 to 10 questions works. For shorter pages (product pages, tools), 5 questions is enough. For a landing page, the FAQ should feel like it belongs, not tacked on. A user should recognize the questions as things they were about to ask.
Should the FAQ answers be short or long?
Short. 40 to 100 words per answer. LLMs prefer answers that are complete in themselves and can be extracted whole.
Answers that reference other parts of the article ("as we discussed above") get discounted because they cannot stand alone. Answers that are 300 words with subheadings inside them are also poorly extracted. Aim for one complete, self-contained thought per answer.
See Google's FAQPage guidelines for their answer content recommendations. Same principles apply for AI extraction.
Does adding FAQ schema hurt my SEO?
No. Google confirmed FAQ schema does not affect rankings directly, but it does not hurt them either. The schema is parsed and stored, but does not surface in SERPs for non-authoritative sites. That means adding FAQ schema is free upside for AEO with zero SEO downside.
See Google Search Central Blog for their ongoing structured data guidance.
What other schema markup should I add alongside FAQ?
Three schemas cover most SaaS use cases:
- Article schema: for every blog post. Include headline, author, datePublished, dateModified.
- FAQPage schema: for the FAQ section.
- HowTo schema: for step-by-step guides.
Product pages should add Product schema with Offer. Comparison pages benefit from ItemList. Validate everything with Schema.org Validator.
What we did at Growth Pigeon
We added FAQ schema and visible FAQ sections to every article and landing page. Different FAQs per page, matched to the specific intent of that page. Combined with question-style H2s and visible "last updated" dates, this stacked structural signal moved our AI citation rate from 0% to approximately 12% on target prompts in 90 days. Read the full pillar on AEO for SaaS Founders for the strategic frame.
FAQ
Does FAQ schema still show up in Google search results?
Only for authoritative government and health sites. For everyone else, Google stopped displaying FAQ rich results in 2023. But the schema itself is still valid and still parsed. LLMs use it as an extraction signal, which is where the current AEO lift comes from.
Can I use FAQ schema on landing pages, not just blog posts?
Yes, and you should. Landing pages benefit heavily from FAQ schema because they typically target commercial-intent queries where citations drive real conversions. Add 5 to 8 questions specific to the offer on the page.
Should I use FAQ schema or HowTo schema for tutorials?
Both. Use HowTo for the step-by-step section of the tutorial. Use FAQ for the questions readers have after reading the tutorial. They are complementary, not redundant.
How do I know if my FAQ schema is valid?
Run the page URL through Google Rich Results Test. If the tool shows a FAQPage detected with all questions and answers, it is valid. If it shows errors, fix them before publishing.
Does the FAQ order matter?
Put the most important question first. LLMs give slight extra weight to the first Q&A in a FAQPage schema block. If you have one question that maps to your highest-priority target prompt, lead with it.
Can I generate FAQs with AI?
You can generate a draft, but you should not publish AI-generated FAQs without review. AI tends to generate generic questions no real user asks. Take the AI draft as a starting point, then rewrite each question in real user language and replace generic answers with specifics from your product or research.
What if I already stopped implementing FAQ schema after Google deprecated it?
Add it back. This is a 30-minute fix per page that measurably lifts AI citations. The technical work is done in your template once, then applies to every article. There is no downside.
What to do this week
Three moves:
- Add FAQ schema to your top 5 highest-traffic articles. Different FAQs per article, matched to that article's intent.
- Update your article template or theme so FAQ schema is emitted automatically for future posts.
- Run every updated page through Google Rich Results Test to verify the schema is valid.
Total time: 4 to 6 hours. Expected lift: measurable within 4 to 8 weeks on Perplexity and AI Overviews. Read how to check if AI cites your site for the measurement workflow, or get a full AEO Audit for a prioritized fix list beyond just FAQ schema.