glossary · faqpage schema
What FAQPage schema marks up, and what Google changed
Structured data for the questions a site asks and answers itself. What the markup contains, why the FAQ rich result mostly disappeared in 2023, and why the block is still worth emitting.
FAQPage schema is JSON-LD structured data that marks up a list of questions and the site's own answers on a page, so search engines and AI assistants can read each pair as a discrete question and answer.
reviewed 2026-09-02 · by the IT Master editorial team · how we check facts
What FAQPage schema is in practice
FAQPage is a Schema.org type, and in practice it is one JSON-LD block on a page that already shows an FAQ. Inside it, mainEntity holds an array of Question items; each question carries a name and an acceptedAnswer whose text is the reply. Nothing else is required.
Three rules decide whether the markup is valid rather than decorative:
- The answers must be on the page. Text that exists only inside the script tag describes a page the reader never sees.
- The site writes both halves. FAQPage is for questions you ask and answer yourself. Where users supply competing answers, as on a forum or support board, the correct type is QAPage.
- One FAQ block per page, on a page whose purpose that FAQ genuinely serves.
The markup adds no information a human did not already have. It restates what is visible in a shape a parser cannot misread: where one question ends, which paragraph answers it, and which text belongs to neither. See JSON-LD for the container format itself.
FAQPage schema adds no new information. It restates the questions and answers already visible on the page in a form no parser can misread, which is why it outlived the rich result it was built for.
What changed, and why it still matters in 2026
In August 2023 Google narrowed FAQ rich results to well-known government and health sites. For everyone else the expandable accordion under the listing stopped appearing and the Search Console report for it went away. Plenty of sites deleted the markup on the spot.
That was an overcorrection. The rich result was one consumer of the data, not the only one. What the block still does:
- It hands any parser a clean list of question and answer pairs, which is the exact unit an assistant quotes.
- It marks where an answer starts and ends for anything that fetches a URL and then has to find the relevant passage inside it, assistant search and the crawlers behind AI Overviews included.
- Other engines set their own display rules, and those are not Google's.
Be honest about the ceiling: none of those engines says whether it reads FAQPage markup, let alone what it weighs, so treat the block as making extraction reliable rather than as a ranking signal. The larger gain is upstream anyway. Writing to this format forces short answers that stand without the paragraph above them, which is the same property that gets a passage lifted. Answer engine optimisation and FAQ discipline overlap almost completely.
How this engine handles FAQ blocks
Every article published through IT Master ends with an FAQ, rendered visibly and emitted as FAQPage JSON-LD from the same data, alongside the rest of the page's JSON-LD. The block at the foot of this page is one.
The questions are not invented to fill a slot. They come from People Also Ask and related-query data pulled during discovery, from the site's own Search Console queries once it is connected, and, where it exists, from the support questions and product Q&A in the customer's first-party data. Answers are written to stand alone, then face the same gauntlet as the rest of the draft: fact-check and novelty judged by models from a different vendor than the writer, up to three revision rounds.
The honest limit: for a typical commercial site this markup will not produce an FAQ rich result in Google, and we do not pretend otherwise. It is emitted because it costs nothing at generation time, keeps the pairs machine-readable for everything else that reads the page, and because an answer written to stand on its own improves the article whether or not a parser ever uses the block. How it works covers the rest of the pipeline.
Five common misunderstandings
"Adding FAQ schema gets you a rich result." Not since August 2023, for most sites. If that was the only reason to add it, the reason has gone; what remains is extraction and structure.
"More questions is better." Ten padded questions is a worse page than three real ones. The markup describes content, it does not improve it, and questions nobody asks read as filler to a reader and to a judge model alike.
"The answers can sit in a tab or an accordion." They can, provided the text is in the HTML the crawler receives and a reader can reach it. What breaks the rule is text present only in the JSON-LD.
"FAQPage and QAPage are interchangeable." They are not. QAPage describes a page where users post competing answers and vote on them. Using it for your own copy misdescribes the page, and inaccurate structured data is worth less than none.
"The markup is a ranking factor." No engine has said so. Structured data helps machines parse a page. The ranking work is still done by the answer.
questions people ask
Is FAQPage schema still worth adding in 2026?
Yes, though for different reasons than in 2021. Google restricted FAQ rich results to well-known government and health sites in August 2023, so a typical commercial page will not get the expandable listing any more. What the markup still does is hand every other parser a clean list of question and answer pairs: assistants that fetch the URL, other engines with their own display rules, and any internal tooling that reads your pages. It costs a few lines of JSON-LD. Add it where a real FAQ already exists, and do not bolt on an FAQ purely to have something to mark up.
What is the difference between FAQPage and QAPage schema?
FAQPage is for questions a site asks and answers itself, with one authoritative answer each. QAPage is for a page where users ask and other users answer, usually with several competing replies, votes, and an accepted answer chosen by the community. Forums, support boards and community threads are QAPage. Your own help section, product FAQ or article FAQ is FAQPage. Choosing the wrong one describes the page inaccurately, and inaccurate structured data is worth less than none, because whatever reads it now holds a wrong model of what the page contains.
Can I mark up answers hidden behind an accordion?
Yes, provided the text is in the HTML the crawler receives and a person can open it. Collapsed accordions and tabs are fine; content injected only after a click that triggers a fetch may not be seen. What is never allowed is text that exists inside the JSON-LD and nowhere on the visible page. Google's structured data guidelines treat that as markup that misrepresents the page, which risks a manual action rather than any benefit. The safe test takes a minute: view source, search for the answer text, then confirm a reader could reach it.
How many questions should an FAQ block have?
As many as there are real questions, which is usually three to six. The number that helps is the number a reader would actually ask before buying or acting, and those come from data rather than imagination: People Also Ask boxes, the queries already bringing people to the page in Search Console, and the questions a support inbox answers every week. Padding the block to hit a target adds words that a reader and a judge both recognise as filler. If you cannot write a standalone answer worth sixty words, the question does not belong there.
See what search engines and AI assistants find on your site
Free, no account. Type your address and we show you what is missing and what we would write first.