Why Your FAQ Isn't Getting Quoted

An FAQ block fails to get quoted when its answers cannot stand alone — which is usually because they were written for someone who has already read the page. FAQ sections are the most citation-ready format on the web and the most consistently wasted, because the format is copied without the constraints that make it work.

Below are the six patterns that account for nearly all of it, each with a before-and-after and a test.

Pattern 1: The answer starts with "Yes" or "No" and nothing else

A one-word opening carries no information once the passage is separated from its question. An extraction system may take the answer without the heading, and "Yes, absolutely — and it's easier than most people think" is a chunk that says nothing.

Before: Q: Does AEO replace traditional SEO? — A: No, not at all. They work together.

After: A: AEO does not replace traditional SEO. The retrieval gates run in order, and crawlability and parsing are conventional technical SEO work that must succeed before extraction is possible.

Answer with the polarity and the restated subject. "No" followed by the full proposition costs you eight words and makes the passage self-sufficient.

Pattern 2: The question is not one anyone asks

FAQ questions written backwards from a keyword rather than forwards from a user produce phrasing no human types. "What are the benefits of AEO services for enterprise organisations?" is a heading, not a question.

The test: read the question aloud. If you would not say it to a colleague, a user did not type it into a prompt box.

Source real questions from sales call recordings, support tickets, the "People also ask" box, and the follow-up questions your own AI assistants suggest. Those are phrasings that exist in the wild and therefore have embeddings close to real queries.

Pattern 3: The answer depends on the page around it

The most common failure, and the hardest to see when you are the author. An answer that says "as covered above," "this framework," or "the third option" is unusable once lifted.

Before: A: It depends on which of the three approaches you chose earlier. The second one generally works better for larger catalogs.

After: A: For catalogs above roughly 10,000 SKUs, indexing only filter combinations with demonstrated search demand outperforms indexing all of them, because it concentrates crawl budget on pages that can rank.

The rewrite names the threshold, the choice, and the reason. Nothing outside the paragraph is required to understand it.

Pattern 4: The answer is too long to be an answer

An FAQ answer that runs 400 words is an article section wearing an FAQ's clothes. Extraction systems favour passages that resolve a question completely and briefly — roughly 40 to 120 words is the range that reads as an answer rather than a discussion.

If an answer genuinely needs more room, that is a signal it deserves its own section or its own article, with the FAQ carrying a short answer and a link to the longer treatment.

Pattern 5: The FAQ exists only in schema

Some teams add FAQPage markup describing questions and answers that are not visible on the rendered page, on the theory that the markup is what machines read.

This is a contradiction between structured claims and rendered content, and it is the kind of mismatch a trust-weighted system is specifically built to catch. Google's structured data guidelines require FAQ content to be visible to users; markup describing invisible content is a violation, not a shortcut.

The correct order is always: write the visible FAQ, then mark up what is there.

Pattern 6: Every answer ends in a pitch

An answer that resolves into "— and that's exactly what our team helps with" trains extraction systems to treat the passage as promotional rather than informational. Promotional passages get retrieved less and cited less.

Keep the answer clean. Put the commercial move in a separate CTA block after the FAQ, where it belongs and where it does not contaminate a quotable passage.

The extraction test

One procedure catches all six patterns. Copy a single answer, paste it into a blank document with no question above it, and read it.

Three questions about what you see. Does it state its own subject, or does it assume one? Can you tell what claim is being made without the question? Would it be useful to someone who arrived at this paragraph and nothing else?

If any answer is no, the pair fails. This test takes about fifteen seconds per answer and finds more real problems than any tool.

Markup that matches

Once the visible FAQ is right, the markup is mechanical. Each pair maps to one Question with one acceptedAnswer, and the text in the markup must match the text on the page.

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [{
    "@type": "Question",
    "name": "Does AEO replace traditional SEO?",
    "acceptedAnswer": {
      "@type": "Answer",
      "text": "AEO does not replace traditional SEO. The retrieval gates run in order, and crawlability and parsing are conventional technical SEO work that must succeed before extraction is possible."
    }
  }]
}

Two rules worth holding to. Mark up only questions genuinely visible on the page. And keep the text value identical to the rendered answer — paraphrasing between markup and page reintroduces the mismatch problem the markup was meant to solve.

Frequently asked questions

How many questions should an FAQ section have?

Between four and eight for a typical article. Beyond that, questions tend to become padding, and a page with thirty marginal questions dilutes the strong ones rather than adding coverage.

Can FAQ schema alone get my page into AI answers?

No. Schema describes structure that already exists; it cannot make an unextractable answer extractable. A visible, self-contained FAQ with no markup outperforms invisible markup with no visible FAQ.

Should FAQ answers repeat information from the article body?

Some repetition is correct. A chunk is retrieved independently of the page it came from, so an FAQ answer that assumes the reader saw the body will not stand alone. Restating a definition briefly is a feature, not redundancy.

Does it matter where the FAQ sits on the page?

Placement matters less than self-containment, but near the end is conventional and works well — it lets the body carry the argument and the FAQ carry the short, liftable resolutions.

Where to start

Take the FAQ on your highest-value page and run the extraction test on every answer. In most cases a majority fail on pattern 3, and the fix is a single rewritten opening sentence per answer — perhaps twenty minutes of work for the whole block.

Our free diagnostic tools cover the crawler and directive layer underneath. For a review of how a full site performs across the retrieval path, that is what a strategy session is for.