Decide page by page. Keep FAQPage markup that matches questions already visible on the page, add it where a real FAQ section exists and someone can keep the answers current, and skip it when the only goal is a Google FAQ rich result.
FAQPage is still valid Schema.org markup. Google stopped showing FAQ rich results in Search in May 2026, so the markup no longer earns that display there. It still describes your questions and answers in a standard, machine-readable form, and other systems read Schema.org markup under their own rules.
Start with one page. Compare each marked-up question and answer with the text a visitor can read, check what your theme or plugin already outputs, then validate the JSON-LD. A passing test never guarantees a rich result, a ranking, traffic or an AI citation.
What changed in Google Search
Google's history with FAQ results explains why so many pages still carry the markup. In August 2023, Google announced changes to HowTo and FAQ rich results: FAQ rich results would be shown only for well-known, authoritative government and health websites, and other sites would not see them regularly. Google also said sites did not need to remove the markup proactively.
The next step came in 2026. Google's documentation update log records a notice dated May 8, 2026, saying FAQ rich results would stop appearing in Google Search starting May 7, 2026. On June 15, 2026, Google removed its FAQ rich-result documentation because the feature was no longer shown. Our news post on the FAQ retirement covers the announcement itself.
The display is something you can observe: the FAQ dropdowns are gone from Google results. The markup is a separate matter. Schema.org defines FAQPage as a WebPage that presents one or more frequently asked questions, and that definition did not change when Google retired its display. A vocabulary type and a search feature are different things, and each provider decides how it uses the vocabulary.
Keep, add or skip FAQPage markup
Run each page through the same three-way test. The answer can differ between a help center, a product page with a few questions and a forum thread, even on one site.
Keep existing markup when the page is a genuine FAQ page, every marked-up question and answer is visible, and the answers are current and useful. Keep it too when you want the content available in machine-readable form for several consumers, and when the existing markup is accurate and does not create a duplicate or conflicting graph. Google's own 2023 note said sites did not need to remove the markup, and the 2026 retirement of the display is not by itself a reason to delete accurate markup.
Add markup when the page already contains a clear FAQ section whose questions reflect real customer or reader concerns. The other condition is upkeep. If your team cannot keep the answers current, the markup will drift away from the page and become inaccurate, which is worse than having none. Also confirm that your platform supports adding JSON-LD safely, as covered in the platform section below.
Skip markup in these cases:
- The only goal is a Google FAQ rich result, since Google no longer shows one.
- The FAQ is hidden, duplicated, or generated automatically without editorial review.
- The page is a single user question with several community answers. QAPage fits that shape better.
- The answers contain claims you cannot support, or the page has no real FAQ content.
Write the FAQPage JSON-LD
Each question on the page becomes one Question object inside mainEntity, and each question gets one acceptedAnswer with an Answer object whose text holds the answer. Schema.org Question and Schema.org Answer define those pieces. The example below uses a fictional shop on the reserved example.com domain.
{
"@context": "https://schema.org",
"@type": "FAQPage",
"@id": "https://example.com/help/faq/#faq",
"url": "https://example.com/help/faq/",
"name": "Example FAQ",
"mainEntity": [
{
"@type": "Question",
"name": "Does Example Garden ship nationwide?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes. Example Garden ships standard orders to all 50 U.S. states."
}
},
{
"@type": "Question",
"name": "How long does standard shipping take?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Standard shipping usually takes three to five business days after the order is processed."
}
}
]
}Replace the @id and url values with your own FAQ page address, and give name the title of your page. Then swap each question and answer for the exact wording a visitor reads on the page. Add one more Question object, separated by a comma, for every extra question, and leave no comma after the last one. JSON does not allow comments or trailing commas, so a stray one breaks the whole block.
To publish it, wrap the same JSON in a script element in the page's HTML. The element holds raw JSON-LD, and the visible questions stay in the page body.
<h2>Does Example Garden ship nationwide?</h2>
<p>Yes. Example Garden ships standard orders to all 50 U.S. states.</p>
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "Does Example Garden ship nationwide?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes. Example Garden ships standard orders to all 50 U.S. states."
}
}
]
}
</script>If you would rather not write the JSON by hand, the Schema Nova FAQ schema generator builds the same structure from the questions and answers you enter. Replace its example values with the wording from your page, because the output is only as accurate as the text you give it.
Match the markup to the visible page
Google's general structured data guidelines ask for content that is representative, visible, accessible and not misleading, and that standard is a sound rule for every consumer of your markup. Each question in the JSON-LD should appear on the page, and each answer should say what the JSON-LD says. Wix's structured data troubleshooting page states the same visible-content requirement for markup added on that platform.
Take a clinic front desk that changes its appointment answers when opening hours change. The page text is edited in the page editor, and the JSON-LD sits in a separate field, so one is easily updated without the other. Treat the pair as a single edit: change the visible answer, then change the matching text value in the same sitting.

Keep one answer per question. An ordinary FAQ page should not carry several competing answers to the same question, because that is the shape of a community thread rather than a company's own answers. When one user question has several user-submitted answers, use QAPage instead. For a company page answering ten common customer questions, FAQPage is the better fit.
Longer answers are fine as long as the text value carries what the page carries. Let the number of questions follow what your visitors actually ask rather than a target count. Padding the section with invented questions only adds more text to keep accurate.
Check what your theme, plugin or app already outputs
Before adding a block, view the rendered source of the page and search for FAQPage. Many themes, SEO plugins, page builders and apps already output FAQ markup when you use their FAQ block. Adding a second block on top creates two descriptions of one section, and they can disagree after the next edit.
If markup already exists and matches the page, leave it in place and spend the time on accuracy instead. If it is wrong, fix it at its source: edit the plugin's FAQ block, the theme setting or the app field that generates it, rather than layering a second script over it. Our guide to schema markup on WordPress walks through finding existing output and choosing an install route, and how to add schema markup covers other platforms.
Platform behavior varies, so confirm it in current documentation. Shopify's help center says supported themes include structured data automatically for many content types, and whether an FAQ section is included depends on the theme or app. On a Next.js site, the JSON-LD guide recommends rendering the script in a page or layout component and sanitizing the serialized content. On Squarespace, Webflow, HubSpot and custom builds, the practical route is one JSON-LD script that carries the visible FAQ content, and the platform's own help pages say where code can go.
Make the FAQ readable for Bing and AI search
Bing reads Schema.org JSON-LD, Microdata and RDFa, so the same block serves it, and its URL Inspection tool shows the markup Bing recognizes on a page. Yandex offers a structured data validator for the same formats.

For ChatGPT Search, allow OAI-SearchBot in robots.txt; OpenAI says placement is not guaranteed (ChatGPT Search guidance, crawler documentation). For Perplexity, allow PerplexityBot (crawler documentation). Keep the FAQ text clear in the visible page for all of them, and see whether AI systems use schema markup.
Test the markup and inspect the rendered page
Paste the adapted JSON-LD into the Schema Markup Validator, which checks generic Schema.org vocabulary. A clean result means the syntax parses and the types and properties exist in the vocabulary. It does not say anything about Google eligibility, because there is no Google FAQ feature left to be eligible for. The Schema Nova schema validator is another place to check JSON syntax and structure.
Google's Rich Results Test is still useful for general Google structured-data diagnostics on the other markup on your page. Expect it to report no FAQ result, and do not treat that as an error in your markup. Bing users can open URL Inspection and read the Markup card, and Yandex provides the validator linked above.
Then look at what the live page actually serves. Open the page source after publishing, find the script element and confirm that the questions and answers inside it still match the visible text, word for word. Single-page apps and plugins that inject code after load can output something different from what your editor shows. The Schema Nova schema checker can retrieve a page address and report the JSON-LD in the fetched HTML.
Recheck after every content or platform change. A theme update, a plugin swap or an edited answer can leave the markup out of step with the page.
Questions people ask
Does FAQ schema still work in 2026?
It is still valid Schema.org markup, and it still describes your questions and answers in a machine-readable form. It no longer produces FAQ rich results in Google Search. Other systems read Schema.org under their own documentation, so check each provider before expecting a display.
Should I remove FAQ schema from my website?
Not just because Google stopped showing FAQ results. Google said in 2023 that sites did not need to remove the markup proactively. Remove it when it is inaccurate, hidden from visitors, duplicated by another block, or impossible for your team to keep current.
Can Bing, ChatGPT Search or Perplexity use FAQ content?
Bing says it reads Schema.org JSON-LD. OpenAI and Perplexity document crawlers for their search products, and OpenAI says placement is not guaranteed. Allow those crawlers if you want eligibility, and keep the FAQ text itself clear in the visible page.
What is the difference between FAQPage and QAPage?
FAQPage describes a page where the site publishes its own answers to common questions. QAPage fits a page with one user question and one or more user-submitted answers, such as a forum thread. Match the type to the real shape of the page.
Does FAQ schema improve rankings?
Google says structured data makes a page eligible for supported features but does not guarantee an appearance, and a passing test is no promise of a ranking. Write the markup to describe the page accurately and treat any benefit as a possibility, not a result.
