Quick Answer: Only if you already publish real question-and-answer pairs: Google stopped showing FAQ rich results for every site on 7 May 2026. FAQPage remains a valid schema.org type that labels each question and answer verbatim, which costs nothing and does no harm. The visible pairs, not the JSON-LD, are what AI engines quote.
FAQ schema had one clear payoff for years: an expandable question list under your Google listing. That payoff is gone, and Google has deleted the documentation page that described it. What remains is a labelling layer, and whether it is worth keeping depends on what you were using it for.
This page is about that one schema type and that one decision. For the wider question of which structured data types still matter for AI engines, see schema markup for AI search.
What Google actually retired and when
Google retired the FAQ rich result, the visible search feature, in two steps: a restriction in August 2023 and a full shutdown on 7 May 2026. The FAQPage schema type itself was never withdrawn; only Google's use of it in results was.
The 2023 restriction
On 8 August 2023 Google announced that FAQ rich results "will only be shown for well-known, authoritative government and health websites. For all other sites, this rich result will no longer be shown regularly." The same post added that "there's no need to proactively remove it. Structured data that's not being used does not cause problems for Search, but also has no visible effects in Google Search."
That wording is why the markup lingered on so many pages. It was harmless, the plugins kept generating it, and a handful of government and health sites still saw the feature.
The 2026 shutdown
Google's changelog entry of 8 May 2026 states that the feature stopped appearing in Google Search on 7 May 2026. Its exact wording is "This feature will no longer appear in Google Search starting May 7, 2026", and on 15 June 2026 the changelog recorded that the FAQ documentation was removed entirely, because "the FAQ rich result feature is no longer shown in Google Search results."
The old documentation URL now redirects to that changelog entry. As of 16 September 2026, there is no Google page describing how to mark up an FAQ for Search, because there is nothing in Search that consumes it.
The timeline in one table
| Date | What changed | Source |
|---|---|---|
| 8 August 2023 | FAQ rich results limited to well-known, authoritative government and health sites | Search Central blog |
| 7 May 2026 | FAQ rich results stop appearing in Google Search for all sites | Changelog, 8 May 2026 |
| 15 June 2026 | FAQPage documentation removed; URL redirects to the changelog | Changelog, 15 June 2026 |
What FAQ schema still does for AI search
FAQ schema still labels each question and its answer as a machine-readable pair, and that is the whole of what it does. No AI engine has published a statement that it reads FAQPage markup specifically, and Google says its AI features need no special schema at all.
What Google says about schema and AI features
Google's AI features documentation states: "You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add." The one structured-data instruction it gives is to make sure "your structured data matches the visible text on the page."
Its AI optimization guide files "overfocusing on structured data" under things you can ignore, while adding that it is "a good idea to continue using it as part of your overall SEO strategy." Read together, Google's position is that structured data is neither required for nor rewarded by AI Overviews and AI Mode. That applies to ranking in AI Overviews exactly as it applies to classic results.
What the other engines say
OpenAI's, Anthropic's and Perplexity's crawler documentation describes fetching and indexing pages and says nothing about FAQPage or any other schema type. Their fetchers read the rendered HTML, and a well-formed question heading followed by a direct answer is what they extract from it.
That is the honest case for FAQ content, as opposed to FAQ schema. The pairs get quoted because the answer sits in the first sentence under a question the user actually typed, which is the pattern behind structuring content for AI extraction. The JSON-LD is a mirror of that pattern, not the source of it.
Why keeping it is still reasonable
Three reasons, none of them a ranking claim. FAQPage is still a valid schema.org type, defined as "a WebPage presenting one or more Frequently asked questions," so validators will not flag it, and Google has said unused structured data causes no problems.
The third reason is discipline. Maintaining the block forces you to write each answer as a self-contained 40 to 60 word unit, which is the habit that gets a page cited.
When to keep FAQ schema and when to drop it
Keep FAQ schema when the page has a genuine visible FAQ; drop it when the block was generated to chase a rich result that no longer exists. The decision turns on the visible content, not on the markup.
| Situation | Decision | Why |
|---|---|---|
| Page has 5 to 8 visible Q&A pairs that answer buyer questions | Keep | The markup mirrors real content and costs nothing |
| FAQ block was added in 2019 to 2023 for the rich result only | Remove or rewrite | A hollow FAQ adds page weight and thin text |
| Answers in the JSON-LD differ from the visible text | Fix or remove | Google asks that structured data match visible text |
| Plugin auto-generates FAQPage from headings | Turn it off | Headings are not questions and the pairs will be wrong |
| Government or health site that had the rich result until May 2026 | Keep | Nothing to gain by removing it; the content is still read |
The plugin trap
Several CMS plugins generate FAQPage from any accordion or toggle block. If the "questions" are navigation labels or product features, the markup describes pairs that do not exist. Since Google no longer surfaces the feature, nothing catches the mismatch except a manual check, so audit the output once and disable auto-generation where it is wrong.
What to do with the freed effort
The hours previously spent on FAQ markup belong in the answers themselves. A question heading, a first sentence that answers it in under 30 words, and one fact per pair is what gets a page cited by ChatGPT; it also reads better to a person than an accordion.
FAQ schema generators and validators
A generator saves typing but produces the same 20-line structure shown above, so the value is in the validation step rather than the generation step. Two tools were checked from their own pages on 16 September 2026.
Generators
Rank Math's WordPress plugin adds an "FAQ by Rank Math" block to the block editor. Its own documentation now notes that Google "used to display" the FAQ rich snippet, with a screenshot of how it "used to look."
Any CMS block that renders the visible pairs and emits matching JSON-LD does the same job. For a static site, a hand-written block in the page template is the most reliable option, because nothing can silently regenerate it.
Validators
The Schema Markup Validator at validator.schema.org checks that the JSON-LD parses and that the types and properties exist in the vocabulary. It does not check whether the text matches the page, so pair it with a manual read of the rendered HTML. Confirm the block is present in the server-rendered source, because a block injected by client-side JavaScript may never be seen by a fetcher that does not execute scripts.
How to write the FAQ pairs that get quoted
The pairs that get quoted are the ones where the question matches a real query and the first sentence answers it outright. Everything else about the format is secondary, and the markup is irrelevant to it.
The shape of a quotable pair
Write the question as the user would type it into an AI engine, in the second person and with the year where the answer changes yearly. Answer in the first sentence, in under 30 words, then add one supporting fact and stop. Each pair should carry a fact the body of the page did not already state, so the FAQ extends the page rather than summarising it.
How to check whether they are working
Ask the question in Google AI Mode, ChatGPT and Perplexity with browsing on, and note whether the answer carries your wording or your URL. That is one line in an AI visibility audit, and FAQ pairs are the cheapest place to test because each is a single query.
Pages whose pairs never surface need better answers, not better markup. That is the lesson the whole of answer engine optimization keeps teaching.
Frequently Asked Questions
What is FAQ schema?
FAQ schema is the schema.org FAQPage type, a WebPage that presents one or more frequently asked questions. It is written as JSON-LD with a mainEntity array of Question items, each holding an acceptedAnswer. Google used it to show an expandable question list under a search listing until 7 May 2026.
Is FAQ schema still relevant in 2026?
Only as a labelling layer for content you already publish. Google stopped showing FAQ rich results for all sites on 7 May 2026 and removed the documentation on 15 June 2026. The markup remains valid, does no harm, and forces disciplined question-and-answer writing, but it earns no search feature and no engine has said it reads it.
Did Google penalise sites that kept FAQ schema?
No. Google's August 2023 post said there was no need to proactively remove the markup, and that unused structured data does not cause problems for Search. Nothing in the May or June 2026 changelog entries changed that. The only risk is markup that contradicts the visible text, which is a general structured-data rule, not an FAQ-specific one.
Does FAQ schema help with AI Overviews or AI Mode?
Not directly. Google's AI features documentation says there is no special schema.org structured data you need to add, and its AI optimization guide lists overfocusing on structured data among things to ignore. Pages appear in those features by being indexed and eligible for a snippet; the visible answer text is what gets used.
Does ChatGPT or Perplexity read FAQPage markup?
Neither company has published a statement that it parses FAQPage. Their crawler documentation describes fetching and indexing pages, with no mention of schema types. In practice their fetchers read the rendered HTML, so a question heading followed by a direct answer is extracted whether or not JSON-LD sits beside it.
How many questions should an FAQ section have?
Six to eight is a practical range: enough to cover the buyer questions the body did not answer, few enough that each answer stays specific. Keep every answer between 40 and 60 words with the answer in the first sentence. Longer lists tend to repeat the body, which gives an engine nothing new to quote.
Can I use FAQ schema on a product or service page?
Yes. FAQPage is a WebPage subtype and can sit on any page with visible question-and-answer content, including product, pricing and service pages. The same rules apply: the pairs must be visible, the text must match the markup, and one FAQPage block per page. Since May 2026 the placement changes nothing in Google's results.