FAQ content without FAQPage schema

The check looks for an FAQ block in the heading outline: the longest run of immediately consecutive headings at the same level that all end in a question mark. When such a run is at least 3 long and the page declares no FAQPage type, the finding is raised. The threshold lives in config/thresholds.json under rule_engine.SCH-03 as min_consecutive_question_headings_for_faq: 3 and is marked "_status": "PROPOSAL". It is our proposal; the product specification sets no number.

Since 19 September 2026 the FAQPage type is looked for in both formats we read: JSON-LD and schema.org microdata. The format does not decide it. We do not read RDFa, and that is an admitted gap.

In other words: the check guesses that the page looks like an FAQ and asks why it is not labelled as one.

What the check does not do:

• It cannot recognise a question without a question mark. “How long delivery takes” does not count, although it is as much a question as any. The check looks for a character, not for meaning.
• It does not check that an answer follows the question. A page with three consecutive question headings and not one answer below them gets the finding too. We judge by the shape of the outline, not by what sits under it.
• It does not judge the contents of an existing FAQPage. If the type is present, the check goes quiet, even if it contains none of the questions on the page.
• It fires where there is no FAQ at all. An article whose subheadings are posed as questions (which STR-02 positively encourages) will get this finding even though it is no FAQ. These two checks pull against each other at this point, and it is fair to admit it. Since 20 September 2026 it at least takes three questions in a row at the same level, so two question marks scattered across a marketing page no longer raise it.

The finding is page-level; severity is warning.

One thing has to come before everything else here, because it is the main reason anybody ever bothered with FAQPage: Google discontinued FAQ rich results. The deprecation notice is dated 7 May 2026, and since then the expandable questions no longer appear under results, save for verified government and health sites. The Search Console report and the testing tool went in June 2026, API support in August 2026.

So: FAQPage will not get you expandable questions in search, and we are not going to promise you any. Anyone who does is selling you a discontinued feature.

What remains true:

• The FAQPage type is still a valid part of schema.org and Google still parses it to understand the page. It just no longer turns it into a visual element in results.
• It is a valid machine-readable description of the fact that the page holds question–answer pairs. Whether AI platforms actually read it, though, we do not know: Perplexity does not mention structured data in its own crawler documentation, OpenAI and Anthropic do not declare reading it, and for Bing we could not find a source. This page claimed otherwise until 13 September 2026; we had no evidence for it.

And what is not documented: that FAQPage improves your chances of being cited in an AI answer. Everything from SCH-01 applies: measurements within noise, a significant decline for AI Overviews, and above all the most practical finding in the field, information placed only in schema was used by none of the platforms tested.

In practice this yields a single recommendation, and it matters more than the schema: write the questions and answers into the page text as ordinary headings and paragraphs. That is where models will reach them. Add the schema on top afterwards, because it costs nothing and does no harm, not because it is the important part.

A note on this check's role in the audit: it reports that a page looks like an FAQ and is not labelled as one. Whether to label it is your decision, and declining is legitimate, especially for an article with question-shaped subheadings rather than a genuine FAQ.

First decide what the page really is:

1. A genuine FAQ: question and answer pairs where each answer stands on its own. FAQPage belongs here.
2. An article or guide with question-shaped subheadings: the text flows and the questions are how it is divided. FAQPage does not belong here and you may safely ignore the finding.

For the first case the minimal markup looks like this:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "How long does delivery take?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Orders placed before 2pm ship the same day and arrive within two working days."
      }
    },
    {
      "@type": "Question",
      "name": "Can I return an item?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Within 14 days, no reason needed; refunds are processed within three working days."
      }
    }
  ]
}
</script>

Conditions that must hold:

• Both question and answer must be visible on the page. Schema carrying answers that are not in the text is hidden content under Google's policies. And by the measurements, nobody uses it anyway.
• The answer text must be complete. An answer truncated with “…read more” makes sense neither to a person nor to a parser.
• Do not use FAQPage for a page with one question, nor for a forum where users answer. Other types exist for those.
• Do not use FAQPage on a product page as a workaround. Google used to track that as rich-result abuse, and today there is not even a motive.

If you take one thing from this page, take this: the value lies in having a clearly posed question and a direct answer in your text. Schema is the label on the box, not the contents.