FAQ schema tells AI tools like ChatGPT, Perplexity, and Google AI Overviews exactly which questions your page answers and how. Adding it correctly makes your content easier to extract than unstructured prose. But most guides stop at the code block and skip the parts that actually determine whether AI tools will use your answers: question selection, answer writing, and ongoing maintenance.
Google's original August 2023 announcement limited FAQ rich results to authoritative government and health sites for most queries. The change is documented on the Google Search Central Blog under "Changes to HowTo and FAQ rich results." If you're following a guide that promises FAQ snippets in search results, that guide is out of date.
The reason to add FAQ schema in 2026 has shifted. AI tools parse structured data when generating answers. A page with valid FAQPage schema signals that the publisher has identified specific questions and committed to specific answers. That's a useful signal for an AI system deciding which content to extract and cite.
This doesn't mean schema is a citation guarantee. It's one signal among several, including domain authority, content quality, and how directly your page answers a query. But it's a signal most competitors haven't implemented correctly, and in competitive categories, that gap matters. As AI-generated answers become a larger share of how people find information, the case for structured data keeps getting stronger.
The questions you mark up in FAQ schema should mirror the queries AI tools are already generating answers for. Your internal FAQ list is a starting point. It needs refinement based on how AI tools phrase queries in your category.
Use Google's "People Also Ask" boxes, Perplexity's query suggestions, and ChatGPT's autocomplete behavior to find the exact phrasing AI tools use for your category. These phrasings are often different from what your team would write internally.
Prioritize questions with a single, direct answer. "What is your pricing?" is too open-ended for AI extraction. "Do you offer month-to-month contracts?" has an extractable yes/no answer with a follow-on sentence.
For service businesses (contractors, agencies, law firms, medical practices), geo-targeted questions hold up well. "Are you licensed to work in Michigan?" or "Do you serve clients outside of Grand Rapids?" are strong candidates because local AI queries include geographic qualifiers.
A few rules to follow:
Aim for 3 to 8 questions per page. More than 8 dilutes the signal and risks a Google quality flag.
Every marked-up question must appear as visible content on the page. Google's guidelines require this. FAQ schema that exists only in the code is a quality violation that can suppress the schema entirely.
Don't copy the same Q&A pairs across multiple pages. Duplicate schema dilutes specificity and creates a quality signal problem.
AI tools extract answers that are direct, self-contained, and short enough to drop into a response without paraphrasing. Most FAQ pages aren't written this way.
The target format: one declarative sentence that answers the question, followed by one or two sentences of supporting context. Total answer length should stay under 50 words for reliable extraction.
For example, imagine a web design agency with this FAQ answer:
Before: "Great question! At our studio, we believe in transparency, so we offer a range of pricing options depending on your specific needs and goals. Every project is different, and we'd love to chat about yours!"
After: "Our web design projects start at a fixed project fee. We send a detailed proposal after a 30-minute discovery call so you know the full scope before committing."
The second version works for AI extraction because it answers the question without requiring the question to be visible in the same context window. Self-contained answers survive extraction intact.
Avoid hedging language that forces the AI to interpret rather than extract. Phrases like "it depends," "in most cases," and "generally speaking" make the answer harder to use. If the answer genuinely varies, write the most common case as the direct answer and note the exception in a follow-on sentence.
JSON-LD is the implementation format Google recommends for FAQ schema, and it's the format AI crawlers parse most reliably.
Here's a complete, copy-paste-ready example with two question/answer pairs:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "Do you offer month-to-month contracts?",
"acceptedAnswer": {
"@type": "Answer",
"text": "No. Our web design projects use a fixed project fee with a defined scope. You are not locked into a recurring contract."
}
},
{
"@type": "Question",
"name": "How long does a typical project take?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Most projects take 6 to 10 weeks from kickoff to launch, depending on scope and content readiness."
}
}
]
}
</script>
A few notes on the code:
@context always points to schema.org.
@type must be FAQPage. Don't use ItemPage or QAPage for FAQ content.
mainEntity is an array of Question objects. Each Question needs a name (the question text) and an acceptedAnswer with @type Answer and a text property.
The JSON-LD block goes in the <head> or <body> of the page. Google accepts both. Most CMS setups place it in the <head>.
Yoast SEO and RankMath both support FAQ schema through their block editor integrations. Yoast's FAQ block is the most widely installed option: add the block to your page, type in your questions and answers, and the plugin generates the JSON-LD automatically.
For non-WordPress sites or custom builds, the JSON-LD block can be added through Google Tag Manager as a Custom HTML tag fired on specific pages. This is the non-developer path for sites without a plugin ecosystem.
One important step regardless of method: after setup, verify that the plugin is generating FAQPage schema (not a generic schema type) by running the Rich Results Test. Plugin defaults aren't always correct, and the test will tell you what Google sees.
Adding the code and having working schema are two different things.
Run the URL through the Rich Results Test. This confirms whether Google can detect the FAQPage schema and whether it's error-free. A green result means the schema is valid. It does not guarantee a rich result will appear.
Run the URL through the Schema.org Validator as a secondary check. This catches property-level errors the Rich Results Test sometimes misses.
Check Google Search Console under Enhancements > FAQ after one to two weeks of indexing. This shows whether Google has processed the schema and whether any pages have warnings or errors at scale.
Don't validate once and forget about it. When FAQ content changes on the page, the schema markup must be updated to match. Stale schema is a quality violation.
For AI extraction specifically, there's no direct equivalent of the Rich Results Test for ChatGPT or Perplexity.
Yes, but the reason has changed. Google's 2023 update means most sites won't see FAQ rich result snippets in search results anymore. The current case for FAQ schema is AI extraction: tools like Perplexity, ChatGPT, and Google AI Overviews parse structured data when generating answers, and FAQPage schema makes your content easier to extract than unstructured prose.
If your site runs on WordPress, yes. Plugins like Yoast SEO and RankMath add FAQ schema through a block editor interface with no code required. For non-WordPress sites, Google Tag Manager is a code-free path for adding JSON-LD schema to specific pages without editing the site's source files directly.
Three to eight questions per page is the practical range. Fewer than three gives AI tools little to work with. More than eight on a single page risks a Google quality flag if the page's visible content doesn't clearly support each marked-up question. Each question must appear as readable content on the page, not only in the schema code.
FAQ schema improves the odds but doesn't guarantee it. ChatGPT and other AI tools use multiple signals when deciding what to cite, including domain authority, content quality, and how directly a page answers a query. FAQ schema tells AI tools your page has committed to specific questions and answers, which makes extraction easier. It's one signal, not a switch.
Update the schema markup to match the new content. FAQ schema that no longer matches the visible text on the page is a quality violation under Google's structured data guidelines and can suppress the schema entirely. Treat schema updates as part of any content edit workflow, not a one-time setup task.
Published November 2026 • Last reviewed November 2026