FAQ Schema After Google Dropped FAQ Rich Results

FAQ rich results are gone from Google Search. On May 7, 2026, Google stopped showing them, and the rest of the feature was taken apart in stages: the FAQ report in Search Console and FAQ support in the Rich Results Test went in June, Search Console API support went in August, and Google has since removed the FAQ rich result documentation entirely, as recorded on its Search documentation updates page. The expandable questions under a search result are not coming back.

The FAQ section on a page is a separate question, and the answer to that one is still yes. This post explains what changed, what FAQ markup still does and does not do, why a well-written FAQ is more useful in AI search than it was in classic search, and how to write one. It is part of our guide to agent-ready websites.

A short history of FAQ rich results

May 2019. Google launched FAQ and HowTo rich results. Mark up questions and answers with FAQPage, and your search result could grow expandable questions underneath it, taking up far more space on the results page.

2019 to 2023. It became one of the most abused tactics in SEO. Sites added FAQ markup to every page, often with questions nobody asked, written only to push competitors further down the results.

August 2023. Google announced that FAQ rich results would appear only for "well-known, authoritative government and health websites," and deprecated HowTo results on mobile. For almost every business site, the feature was gone from that day.

May to August 2026. Google removed FAQ rich results for the remaining sites on May 7, then removed the Search Console report, the Rich Results Test support, the API data and finally the documentation. Search Engine Journal's coverage has the details of the announcement.

What did not change

Three things are worth stating clearly, because a lot of advice online gets them wrong.

You do not need to remove existing FAQ markup. Unused structured data causes no penalty. Removing it gains nothing, and other systems can still read it.
FAQPage is still a valid Schema.org type. The Schema.org definition is unchanged, and the Schema Markup Validator still checks it.
The general rules still apply. Google's structured data guidelines require marked-up content to be visible to the user. That rule never depended on FAQ rich results.

What changed is the reason most sites added it. The visual reward is gone, and so are the reports that measured it.

Do AI assistants read FAQ markup?

This is where honest answers are scarce, so here is what is actually documented.

Google says in its guidance on AI features and your website that no special markup or new files are needed to appear in AI Overviews or AI Mode, and that structured data you do use should match the visible content.

OpenAI, Anthropic and Perplexity do not say in their crawler or publisher documentation that they use FAQPage markup, or any particular structured data type, when choosing what to cite. They do not say they ignore it either. Claims that FAQ schema "gets you into ChatGPT" are not backed by anything these companies have published.

What is documented is that these systems read the page. Their search crawlers fetch the HTML, and their answers quote text from it. A question written on the page with a direct answer underneath is readable by every one of them, with or without markup.

So the practical position is: the visible FAQ does the work, the markup is a free and harmless description of it. If your platform generates it from the content, keep it. If you would have to maintain it by hand as a separate copy, the visible text matters far more.

Why FAQ sections matter more in AI search, not less

Classic search showed ten links and let the reader do the work. AI search reads the pages and writes the answer. That changes what kind of content is useful.

Assistants answer questions, and an FAQ is a list of answers. When someone asks "does this clinic accept international insurance?", the assistant is looking for a sentence that answers exactly that. A visible question with a direct answer underneath is the clearest match a page can offer.

FAQs state facts plainly. Marketing copy avoids direct statements. An FAQ cannot. "Yes, on weekdays until 20:00." "No, the free plan has no time limit." Assistants find those sentences easily and struggle to infer them from a hero headline.

FAQs fill the gaps that cause wrong answers. When an assistant cannot find a fact on your site, it guesses or takes it from somewhere else: an old directory, a review, a competitor's page. A question that customers ask, answered on your own site in your own words, closes that gap.

Readers still want them. A pricing page that answers "what happens if I cancel?" in two sentences saves a support email. That was true before structured data existed and it is still true.

FAQPage or QAPage?

Schema.org has two similar types, and they are for different situations.

FAQPage is for questions and answers written by the site owner. One official answer per question. This is what a business site uses.

QAPage is for a page where users ask a question and other users answer it, like a forum thread or community Q&A. Google's Q&A documentation describes it, and unlike FAQ it still has a rich result. A business site almost never needs it.

What the markup looks like

A minimal FAQPage block in JSON-LD:

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "Do you accept international health insurance?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Yes. We work directly with Allianz Care, Cigna Global and Bupa Global. For other insurers we provide an itemized invoice you can submit yourself."
      }
    },
    {
      "@type": "Question",
      "name": "Are you open on Saturdays?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Yes, from 10:00 to 15:00. Saturday appointments must be booked in advance."
      }
    }
  ]
}

The text in the markup must match the text on the page, word for word where possible. Anything that reads both will notice a difference.

How to write an FAQ that works

Find the real questions

Good FAQ questions come from evidence, not brainstorming. The best sources:

Support emails and chat logs. The questions people ask more than once.
Sales and booking calls. What people want to know before they commit.
Search Console. The queries in the Performance report that are phrased as questions.
"People also ask" boxes in Google for your main topics.
AI assistants themselves. Ask them about your business and note what they get wrong or cannot answer. Each gap is a candidate question. The method is in How to Audit What ChatGPT Says About Your Business.

Write the answer first

Start with the answer, then add the condition or detail. One fact per answer.

Weak:

What are the benefits of our implant treatment? At Kaya Dental, we pride ourselves on delivering world-class implant solutions tailored to each patient's unique needs, using the latest technology.

Strong:

How long does an implant take from start to finish? Usually three to six months. The implant is placed in one visit, then needs eight to twelve weeks to heal before the crown is fitted. Patients who need a bone graft first should add three to four months.

The second answers a real question with a number, a process and a condition. An assistant can quote it. A person can plan around it.

Keep each FAQ specific to its page

The FAQ on a pricing page answers pricing questions. The FAQ on an implant page answers implant questions. The same block repeated on every page tells readers nothing new and gives an assistant several identical candidates to choose from instead of one clear page that owns the answer.

Keep it short and current

Six to ten questions that matter beat thirty that pad the page. And an FAQ answer with an old price is a wrong answer an assistant will repeat with confidence. When the fact changes, the FAQ changes the same day.

Accordions: what is fine and what is not

There is a lot of confusion here.

Fine: a standard accordion where all the answer text is in the page's HTML and simply collapsed visually. Google indexes content in tabs and accordions normally under mobile-first indexing, and crawlers that read HTML see the full text.

Not fine: answers that are loaded from a server only when someone clicks, so the text is not in the HTML at all. Many AI crawlers do not run JavaScript, so they never see those answers. If the answers exist only in the JSON-LD and not in the visible page, that also breaks the rule that marked-up content must be visible.

To check, view the page source (not the browser inspector) and search for a sentence from one of your answers. If it is there, crawlers can read it.

What to stop doing

Adding FAQ markup for questions nobody asks, to chase a rich result that no longer exists.
Writing the question as a keyword and the answer as a paragraph of marketing.
Answering in a way that avoids the question: "It depends on your needs. Contact us to learn more."
Watching for FAQ data in Search Console reports or API exports. It is no longer there, and dashboards built on it will show nothing.

If you did any of these for rich results, the honest move is to cut the FAQ down to questions a customer would actually ask, rewrite the answers as plain facts, and let the markup describe what is there.

How to check your FAQ

View the page source and confirm every answer is in the HTML.
Run the page through the Schema Markup Validator. The Rich Results Test no longer supports FAQ, so the validator is the tool for confirming the FAQPage block is valid and matches the page.
Read each answer out loud. If it does not answer the question in the first sentence, rewrite it.
Check the facts. Prices, hours, policies and names should match the rest of the site.
Ask an assistant one of the questions. See whether it finds your answer, and whether it cites your page.

How this works in Beste

Add an FAQ section to any page in Beste and the questions and answers are published as FAQPage structured data on publish. The answers are rendered in the page's HTML, so crawlers see them whether the accordion is open or closed. There is no schema plugin and no second copy to keep in sync: edit an answer in the section and the markup updates with it. On a multilingual site, each language's FAQ produces its own markup for that page.

If you keep your site up to date with an AI agent, updating an FAQ answer is one request, as described in Keeping a Website Current With an AI Agent. The SEO-friendly structure page lists the other structured data Beste publishes, and Entity Schema for Small Sites covers the entity data that ties every FAQ answer to the business giving it.

Sources

Checked in October 2026.