How to Write FAQ Pages That AI Answers Quote

To write FAQ pages that AI answers quote, use the real questions buyers ask, answer each one in the first sentence, keep most answers to 40 to 80 words that make sense alone, and publish them as plain HTML. FAQPage schema is optional now: Google stopped showing FAQ rich results in May 2026.

Key takeaways🔗

• AI engines lift single passages. An FAQ answer that names the subject and answers in its first sentence is easy to quote whole.

• Pick questions from sales calls, support tickets, site search and the prompts buyers type into ChatGPT, not from what marketing wants to say.

• Google limited FAQ rich results to well-known government and health sites in August 2023, stopped showing them in Google Search from May 7, 2026, and removed the documentation in June 2026.

• FAQPage markup does no harm and other systems may still read it, but Google says no special schema is needed for its AI features.

• Keep answers in the page HTML, not in accordions that load content by JavaScript.

Why do AI answers quote FAQ pages?🔗

AI answers quote FAQ content because a well-written FAQ is already shaped like the thing the engine is trying to produce: a question, then a short, direct answer. When a buyer asks ChatGPT or Perplexity "does [product] have a free trial," a passage that reads "Yes, [product] has a 7-day free trial with no credit card required" can be used almost as is.

That only works when each answer stands on its own. Engines pull passages out of context, so an answer that starts "Yes, as mentioned above" or "It depends on your plan" with no plan names loses most of its value once it is lifted. Our guide to content strategy for AI citations covers this self-contained style across a whole site; FAQs are the easiest place to start.

FAQs also fill a gap. Product and pricing pages tend to explain what you want to say. FAQs answer what buyers actually want to know, including the uncomfortable parts: cancellation, limits, who the product is not for.

How do you choose which questions to answer?🔗

You choose FAQ questions by collecting the questions real buyers ask before they buy, then keeping the ones that come up often and that you can answer clearly. Good sources, roughly in order of value:

1. Sales calls and demo notes. The questions prospects ask live are the ones they also ask AI.

2. Support tickets and chat logs. Especially pre-purchase questions about limits, setup and billing.

3. Site search queries. What people type into your own search box when they cannot find something.

4. AI engines themselves. Ask ChatGPT, Perplexity and Google AI Overviews the questions in your category and note what they get wrong or leave out about you.

5. Reddit threads and review sites. Read what people ask about you and your competitors in public. Our post on Reddit citations in AI answers explains why these threads matter.

6. People Also Ask boxes in Google for your main terms.

Then filter. Keep a question if it is asked often, if a wrong answer would cost you a sale, or if AI engines currently answer it badly. Drop questions nobody asks ("Why is our company so passionate about quality?") and questions you cannot answer honestly.

Phrase each question the way a buyer would say it, in full: "How much does Acme Payroll cost for 10 employees?" rather than "Pricing?"

How long should an FAQ answer be?🔗

Most FAQ answers should be 40 to 80 words: long enough to answer and give one supporting detail, short enough to be quoted whole. That is a writing guideline, not a published rule from any AI company, and some questions need more. Google's AI features guide says "there's no requirement to break your content into tiny pieces for AI to better understand it" and that "there's no ideal page length."

So aim for complete rather than short. A useful pattern:

• Sentence 1: the direct answer, naming the product or subject. "Yes," "No," a number, or a clear statement.

• Sentence 2 to 3: the condition, limit or detail that makes it accurate.

• Optional: a link to the page with the full detail.

Acme Payroll is made up for this example. The point is the shape: each strong answer could be quoted alone and still be correct.

Should you still use FAQPage schema?🔗

You can keep FAQPage schema, but do not add it expecting a Google feature, because Google no longer shows FAQ rich results at all. Here is the timeline from Google's own records:

The old FAQ structured data page now redirects to that changelog. Google's AI features guide adds that "structured data isn't required for generative AI search, and there's no special schema.org markup you need to add."

So why keep it? The schema.org FAQPage type is a public vocabulary, not a Google product, and other search engines and tools may still read it. Google's 2023 post also said unused structured data "does not cause problems for Search." If your CMS adds FAQPage markup automatically from visible questions, leave it on. On this blog, the FAQ section at the end of each post is turned into FAQPage markup automatically, and TrendlyAI's own pricing page carries FAQPage markup next to its Product and Offer markup. I keep it because it costs nothing and stays in sync with the visible text, not because I expect it to win citations by itself.

Two rules if you use it:

• Mark up only questions and answers visible on the page, word for word.

• Generate it from the same source as the visible FAQ, so the two never drift apart.

For full JSON-LD examples of FAQPage and other types, see our schema markup guide.

Where should FAQ content live on your site?🔗

FAQ content works best close to the topic it answers, with a central FAQ page as a backup. A pricing question answered on the pricing page has context around it; the same answer buried in a 60-question help center is harder to match to the right query.

A practical layout:

• Pricing page: 5 to 8 questions on trials, billing, limits and cancellation.

• Product or feature pages: 3 to 6 questions about what that product does and does not do.

• Comparison pages: questions like "What is the difference between [you] and [competitor]?"

• A central FAQ or help page: everything else, grouped by topic with H2 or H3 headings.

Make sure the answers are in the HTML the server sends. Many FAQ accordions are fine because the text is in the page and only hidden by CSS. Some load the answer by JavaScript only when clicked, and Vercel's study of AI crawler traffic (December 2024) found that "none of the major AI crawlers currently render JavaScript," including those from OpenAI, Anthropic and Perplexity. To check, use "View page source" and search for a phrase from one answer.

What mistakes stop FAQ answers from being quoted?🔗

The most common mistake is writing FAQs as marketing copy instead of answers. Other patterns that make answers hard to use:

• Pronouns instead of names. "It integrates with QuickBooks" loses the product name when lifted. Write "Acme Payroll integrates with QuickBooks."

• Vague hedges. "Pricing depends on your needs" answers nothing. Give a range or the factors that set the price.

• Out-of-date facts. An FAQ that still lists last year's price or a retired plan feeds wrong answers. Date the page and review it when prices or features change.

• Duplicate questions with different answers on different pages. Pick one answer and reuse it.

• Questions nobody asks. Filler questions dilute the page and waste a reader's time.

If you are working through a full site review, the GEO audit checklist includes FAQ coverage alongside the other on-page checks.

How do you know if your FAQ answers are being used?🔗

The direct test is to ask AI engines the same questions your FAQ answers and see whether the response matches your wording or cites your page. Start with the questions where a wrong answer costs you the most, like pricing, trials and key features.

Answers vary from run to run, so a single check is not proof either way. A brand mention tracker repeats the same prompts over time. TrendlyAI does this weekly on ChatGPT, Perplexity, Claude and Google AI Overviews, and can draft an FAQ page with schema when it finds a gap. If you want a quick baseline first, the free AI visibility check runs 5 prompts on ChatGPT and Perplexity in about 30 seconds with no signup. For the wider strategy, start at the GEO hub.

Frequently asked questions🔗

Does Google still show FAQ rich results?🔗

No. Google limited FAQ rich results to well-known government and health websites in August 2023, then stopped showing them in Google Search from May 7, 2026. Google removed the FAQ rich result documentation in June 2026.

Should I remove FAQPage schema from my site?🔗

You do not need to. Google said in 2023 that unused structured data does not cause problems for Search, and other systems may still read schema.org markup. Keep it only if it matches the visible questions and answers exactly.

How many questions should an FAQ page have?🔗

Use as many as buyers really ask, and no filler. On a pricing or product page, 3 to 8 focused questions usually covers it. A central help page can hold more if it is grouped under clear headings.

How long should each FAQ answer be?🔗

Most answers work well at 40 to 80 words, with the direct answer in the first sentence. That is a writing guideline rather than an official limit. Google says there is no ideal page length, so let complex questions run longer.

Do AI engines read FAQ accordions?🔗

They can if the answer text is in the page HTML and only hidden with CSS. If the answer loads by JavaScript when clicked, crawlers that do not run JavaScript, such as OpenAI's and Anthropic's according to Vercel's 2024 study, may never see it.

Related reading🔗

• How to get cited by ChatGPT

• Google AI Overviews optimization

• How to get cited by Perplexity