Update: Since 7 May 2026, Google no longer shows FAQ rich results in its search pages: the feature has been officially retired, even though the FAQPage type remains valid markup you can keep on your pages (Google Search Central, May 2026).
The term is everywhere again since Google cut FAQ rich results, usually with a hasty verdict attached: "FAQ schema is useless now." That skips a step. Below is a clean definition of this markup, what genuinely changed, and why it keeps a role now that generative AI often answers in place of the click.
FAQ schema, defined
FAQ schema, technically the FAQPage type from schema.org, is markup you add to a page's code to tell search engines and AI systems which parts of the content are questions and which are the matching answers. It is structured data: invisible to the reader, it simply makes the question-and-answer structure explicit for machines.
Two words are worth pinning down first. "Schema" refers to schema.org, a shared vocabulary the major search engines built to describe what a piece of content means, in a language machines can read. "FAQ" is a list of frequently asked questions and their answers. FAQ schema, then, is the standardized labelling of a question-and-answer page.
The core idea fits in one sentence: this markup changes nothing about what the visitor sees. It adds, in the code, a description layer that says "this is a question, this is its answer." That is what people mean by structured data, as opposed to the visible text. The reader sees an ordinary page; the machine reads, on top of it, a map of the structure.
An image helps most of my clients get it. Think of a label stuck backstage, behind the page. The visitor never sees it. The engine does. That is the whole principle, and it is far less arcane than the acronyms make it sound.
How the markup works
FAQ schema is usually written in JSON-LD, a small block of code placed in the page. It declares a FAQPage object holding a list of questions, each tied to its answer. Engines read that block to understand, without guessing, that the page is organized as questions and answers.
In practice, the recommended format is JSON-LD: a discreet script tucked into the code that lists each question with its answer. You can also mark up the visible HTML directly, but the principle is the same either way: pair a question with its answer unambiguously. The engine no longer has to interpret the layout to work out where an answer begins; the information is handed to it as is.
This logic reaches far beyond FAQs. The same mechanism describes a product, a review or an event. FAQ schema is just one brick in a larger set of markups that help Google and AI systems map the web. To place this family of tags and its role in AI visibility, we cover it in our guide on structured data for AI.
Markup is only worth something when it tells the truth. FAQ schema must describe questions and answers that are genuinely present on the page. Marking up absent or misleading content breaks Google's guidelines and does nothing for AI systems, which cross-check the code against what they actually read.
The 2026 switch: end of rich results
On 7 May 2026, Google retired FAQ rich results: those expandable question accordions that used to appear under some links no longer show. The FAQPage type is not deleted, though. It stays a valid schema.org markup you can keep without risk.
For years, FAQ schema mostly served one visible goal: winning a rich result, that drop-down box of questions that ate up extra space on the results page. That feature is what Google switched off. Its official update log states plainly that the feature no longer appears in search as of 7 May 2026.
Two things get conflated here, and they should not be. What disappears is the rich display inside Google. What remains is the FAQPage type itself, still valid on the schema.org side. Google notes that unused structured data is not a problem for search: there is no need to rush and strip out existing markup.
Across the 1220 SEO/GEO audits produced by Cicero Studio, structured markup, FAQPage included, is one of the points we check on every site. And I have watched the same mistake go by almost every week since this change: someone tells me "the FAQ tag is dead." No. What died is the display. The tag itself keeps describing the page to machines, exactly as before. It is a crucial nuance, and one that is missed constantly.
What it is still for, on the AI side
The role of FAQ schema shifts from display to comprehension. It helps some engines and systems grasp the question-and-answer structure of a page. On the other hand, no special markup is required to be picked up by Google's AI features: clear, visible content comes first.
The question everyone asks: does marking up your FAQs help you get cited by ChatGPT or by Google's AI Overviews? The honest answer is nuanced. In its official AI optimization guide published in 2026, Google states that no special structured data is needed for its generative features, which lean on its ordinary ranking systems.
Independent studies diverge, and that has to be said clearly. Some associate complete markup with more pickups in AI answers. Others, testing several engines, observe that AI systems extract the visible HTML first and ignore hidden JSON-LD when they fetch a page directly. These are private experiments, to be treated as signals, not laws. The Semrush summary of Google's guide and the Search Engine Journal analysis converge on one point: solid SEO fundamentals remain the bedrock of AI visibility.
The practical conclusion is reassuring: lead with real, clear, visible questions and answers on the page. That is what AI systems read, and what gets cited. FAQ schema comes in support of that content, never in its place. This logic lines up with a FAQ built for AI Overviews and, more broadly, with content designed to get cited by ChatGPT.
We look at exactly that for you, free of charge: where you stand on Google and inside AI answers, and what to reinforce. No factory pitch, just the picture.
Request my free audit →Schema versus content: don't confuse them
FAQ content is the real questions and answers a visitor reads. FAQ schema is the invisible code that describes that structure to machines. The first has value for the reader and for AI systems that read the page; the second is only a description layer that supports the first.
This is the most expensive confusion out there. Plenty of people think "add FAQ schema" and believe they have worked on their content. But markup with no visible questions and answers behind it is worthless: it describes an empty room. Conversely, a genuine section of useful, well-written, visible questions has value even without a single line of markup.
The right way to see it: substance first, label second. You write clear answers to the real questions readers ask, you make them visible on the page, then you optionally add the markup to cleanly document that structure. Never the other way round. A citation by an AI rewards content that answers, not code that promises.
A real question-and-answer section also has a value markup alone will never carry: it captures a crowd of fine-grained intents. According to Cicero Studio's internal data, 35% of the French keywords Cicero Studio analyzed get fewer than 100 monthly searches: the long tail dominates. Across the 4621 French keywords Cicero Studio measured, the median volume is 260 searches per month. Those are precise questions a well-written FAQ can address one by one, exactly where a tag stuck on an empty page captures nothing.
Using it well today
Since the tag stays valid but the display is gone, how should you position yourself? Here is the approach we hold on the ground, without dogma.
- Write real questions and answers first. Start from the questions your customers actually ask, answer them plainly and visibly on the page. That is the true asset, the one both Google and AI systems read.
- Keep existing markup if it is correct. No need to remove it: Google confirms unused markup does no harm. Just check that it faithfully describes the content shown.
- Never mark up an empty room. No invented questions, no answers absent from the page. The markup must mirror reality, or it will be ignored, or penalized.
- Do not expect a display. The FAQ rich result no longer exists in Google. If you add schema, it is for machine readability, not a visual box.
- Mind the overall structure. FAQ schema only makes sense inside a coherent whole: clear headings on content that genuinely answers. It is the whole that becomes quotable.
What FAQ schema does not do
The limits deserve a frank statement, because the false promises around this markup are legion.
The honest limits
- It does not rank a page on its own. Adding FAQ schema does not push a page up the results. It is not a magic ranking factor, it is an aid to comprehension.
- It does not replace good content. Perfect markup laid over hollow answers stays useless. Structure describes value, it does not create it.
- It guarantees no pickup by AI. No code forces ChatGPT or Perplexity to cite you. These systems lean mostly on authority, clarity and what they can verify for themselves.
- It does not lock anything in. Engines re-evaluate constantly, so the real work is the content, maintained, not a tag posted once.
This is also why serious GEO work never reduces to placing tags. The markup is a signal among many, not a passkey. Underneath the acronym, it is still editorial work, made legible to machines and pointed at the questions that matter.
Where Cicero Studio fits
Cicero Studio does not sell markup as a solution in itself. Everything starts with an audit that measures how Google and AI systems actually understand your pages, then editorial production that turns your expertise into real, visible questions and answers, then automated semantic internal linking. Schema, FAQPage included when relevant, comes last, to document what already exists.
So this is not just theory, here is how we put it to work at Cicero Studio, in order. The audit measures where comprehension breaks. Production turns your expertise into questions and answers a reader and a machine can both use. Automated semantic internal linking then wires those pages into a territory. The markup, when it is warranted, arrives at the end to describe cleanly what is genuinely there, never to conjure value that is not.
That is our promise in one line: agency-quality work, software-grade productivity, running as a single loop, GEO audit, editorial production and automated semantic internal linking, rather than three disconnected services. To situate the whole approach, our agence GEO pillar covers it in depth, and our English breakdown of a GEO audit shows exactly what we check.
Resources to go further
We document our approach in the open, because published work with its sources beats any sales deck. Each link below digs into one piece of the answer-visibility puzzle.
Frequently asked questions
What is FAQ schema?
FAQ schema, technically the FAQPage type from schema.org, is markup you add to a page's code to tell search engines and AI systems which parts of the content are questions and which parts are the matching answers. It is structured data: it changes nothing for the reader, it simply makes the question-and-answer structure explicit for machines.
Is FAQ schema still useful in 2026?
It no longer earns rich results in Google search: that feature was retired on 7 May 2026. The FAQPage type itself is still valid markup you can keep, because it helps other engines and systems understand the question-and-answer structure of a page. Its real value shifts from a visual display toward machine readability of the content.
Does FAQ schema improve visibility in ChatGPT or Perplexity?
The evidence is mixed. Google says no special markup is required for its AI features. Several experiments published in 2026 show that AI systems read the visible HTML first and often ignore hidden JSON-LD when they fetch a page directly. Other studies associate complete markup with more citations. The safe bet is to lead with clear, visible questions and answers on the page, with schema as support, never as a substitute.
Should you remove FAQ schema from your pages?
No, that is not necessary. Google states that unused structured data is not a problem for search. You can keep valid FAQPage markup: it remains usable by other services and cleanly documents your page structure. What you must avoid is markup that does not match the content actually shown on the page.
FAQ schema versus visible FAQ content: what is the difference?
Visible FAQ content is the real questions and answers a visitor reads on the page. FAQ schema is the invisible code that describes that structure to machines. The first has value for the reader and for AI systems that read the page; the second is only a description layer. Schema with no matching visible content is useless, and even risky.
Founder of Cicero Studio and an organic-visibility consultant, I help businesses exist on Google as in AI answers. I handle structured markup every week on client sites: it is a useful tool when it documents real content, and a waste of time when you expect miracles from it. The removal of FAQ rich results in 2026 was a good reminder of that limit.
LinkedIn →Sources
- Google Search Central, "Search status and updates" (official update log), May 2026, confirming FAQ rich results were retired as of 7 May 2026.
- Google Search Central, "AI optimization guide" (official documentation), 2026, stating no special structured data is required for Google's generative features.
- Schema.org, "FAQPage" type definition (normative technical reference).
- Search Engine Journal, "Google's new AI search guide calls AEO and GEO still SEO", 2026 (industry analysis).
- Semrush, "Google publishes generative AI search guide", 2026 (industry summary).