FAQ schema is not dead in 2026, but its job changed. Google retired the expandable FAQ rich result on May 7, 2026, so you no longer earn extra SERP real estate. The markup still works: it feeds AI Overviews, ChatGPT Search, and Perplexity, which now read FAQPage data to build answers.
If you stopped at the headline "Google killed FAQ rich results," you got half the story. The visual snippet died. The structured data behind it did not. Those are two different things, and the gap between them is exactly where most SEO advice is now wrong. This guide explains what actually happened, what FAQ schema does today, and how to add it correctly.
What Happened to FAQ Rich Results#
For years, a well-marked FAQPage could earn an accordion of expandable questions directly under your listing in Google search. That dragged extra vertical space onto the page and pushed competitors down. It was one of the highest-leverage pieces of structured data you could ship.
Google started winding this down in 2023, when it limited FAQ rich results to "well-known, authoritative" government and health sites. For everyone else, the snippet quietly disappeared. The final step came on May 7, 2026, when Google fully deprecated the FAQ rich result and stopped rendering the expandable accordion in standard search entirely.
The key distinction: Google deprecated the rich result (the visual feature in the SERP), not FAQPage structured data (the markup in your page's source). The code is still valid schema.org markup, and machines still parse it.
Why the rich result went away#
Google did not explain every reason, but the pattern is clear if you watched the SERP. FAQ snippets got abused. Pages stuffed irrelevant or keyword-stuffed "questions" into the markup purely to grab more space, and the feature degraded into clutter. Google tends to retire features that get gamed at scale, and the FAQ accordion was a textbook case.
The deprecation also fits Google's broader shift toward AI Overviews, where answers get synthesized at the top of the page instead of distributed across ten link snippets. The old FAQ accordion competed with that model. The data underneath it does not, because Google can pull from it directly.
So Is FAQ Schema Worth It in 2026?#
Yes, for most content pages, but for a different reason than 2024. You are no longer adding FAQ schema to win SERP stars or an accordion. You are adding it so answer engines can find, parse, and cite clean question-answer pairs from your page. That is the entire value proposition now, and it is arguably bigger than the old one.
Here is the honest before-and-after:
| Aspect | Before May 2026 | In 2026 |
|---|---|---|
| Expandable FAQ accordion in Google | Yes, for eligible sites | No, deprecated for everyone |
| Extra SERP real estate | Yes | No |
| Read by AI Overviews | Partially | Yes, a primary input |
| Parsed by ChatGPT Search / Perplexity | Yes | Yes |
| Helps machines structure your answers | Yes | Yes, more important than before |
| Risk of penalty for valid use | No | No |
The deprecation removed a vanity benefit and left the durable one. Clean, machine-readable question-answer pairs are exactly what generative engines want, because they map directly onto the way those engines retrieve and quote information.
The new value: structured data for AI search#
Generative engines do not "rank ten links." They retrieve passages, judge relevance, and assemble an answer with citations. Content that is already broken into discrete questions and direct answers is the easiest possible format for that pipeline to consume.
That is why FAQPage markup matters for what people now call GEO (generative engine optimization) or answer engine optimization. Industry analyses through 2026 have observed that pages with FAQPage structured data surface in AI Overviews at meaningfully higher rates than comparable pages without it, with several studies putting the lift around 3.2x. Treat that as directional, not a guarantee, but the direction is consistent: structured Q&A helps you get cited.
You can build valid markup in seconds with a free schema markup generator, which is the fastest way to ship correct FAQPage JSON-LD without hand-writing it.
When You Should (and Should Not) Use FAQ Schema#
FAQ schema is a tool, not a default. Adding it to pages that have no genuine questions is the abuse Google clamped down on. Use it where it fits the content honestly.
Good fits:
- Product, pricing, and service pages with real recurring questions
- How-to guides and tutorials where readers predictably ask follow-ups
- Comparison and "vs" articles where buyers have specific objections
- Support and documentation pages
Poor fits:
- Pages with invented questions written only to trigger markup
- Questions whose answers are not actually on the page
- Duplicating the same FAQ block across dozens of thin pages
- Promotional or ad-like "answers" that do not answer anything
Google's structured data guidelines still require that the question and answer in your markup are visibly present on the page and genuinely answer the question. Invisible or mismatched FAQ markup risks a manual action, even though the rich result is gone.
How to Add FAQ Schema That Works in 2026#
The mechanics are simple. The discipline is in writing questions real people ask and answering them directly. Here is the workflow.
Step 1: Mine the real questions people ask#
Do not invent questions. Pull them from sources where intent is already visible: the "People also ask" box for your target query, the related-questions and autocomplete suggestions in search, your site search logs, and your support inbox. Pick four to eight that genuinely match the page's topic.
Each question should be one a user would actually type or speak. "What is the ideal length?" beats "What about length considerations?" because it matches natural-language queries that AI engines receive.
Step 2: Write tight, self-contained answers#
Answer each question in the first sentence, then add one or two sentences of context. Forty to sixty words per answer is a solid target. Self-contained matters because an answer engine may quote the answer without the question, so it has to stand alone.
Keep the answer text identical to what a visitor sees on the page. The markup mirrors visible content; it does not replace it.
Step 3: Generate valid FAQPage JSON-LD#
Wrap your pairs in FAQPage structured data using JSON-LD, the format Google recommends. A correct block looks like this:
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "Is FAQ schema still worth adding in 2026?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes. Google retired the FAQ rich result, but FAQPage markup still feeds AI Overviews, ChatGPT Search, and Perplexity, which parse it to build cited answers."
}
},
{
"@type": "Question",
"name": "Did FAQ schema get penalized?",
"acceptedAnswer": {
"@type": "Answer",
"text": "No. The rich result was deprecated, not penalized. Valid FAQPage markup that matches visible page content remains safe and useful."
}
}
]
}
Rather than maintain that by hand, paste your questions and answers into the free schema markup generator and let it emit clean, valid JSON-LD you can drop into your page head.
Step 4: Validate and ship#
Run the output through Google's Rich Results Test and the schema.org validator to confirm it parses with no errors. The Rich Results Test will note that FAQ is no longer an eligible rich result, which is expected in 2026 and not a problem. You are validating that the syntax is correct, not chasing a SERP feature.
Then add the JSON-LD to the page, keep it in sync if you edit the visible answers, and move on.
FAQPage vs Other Schema Types for AI Visibility#
FAQPage is one signal among several that help answer engines understand a page. Pairing it with the right supporting types compounds the effect.
| Schema type | What it tells machines | Best for |
|---|---|---|
| FAQPage | Discrete question-answer pairs | Pages with real recurring questions |
| HowTo | Ordered steps for a task | Tutorials and procedures |
| Article / BlogPosting | Author, date, topic, entity | Editorial and blog content |
| Breadcrumb | Site structure and context | Every indexable page |
| Organization | Brand identity and entity | Site-wide, builds entity trust |
Most strong pages stack several of these. An informational guide can carry Article plus FAQPage; a tutorial can carry HowTo plus FAQPage. The goal is to describe the page to a machine as precisely as you can.
If you are optimizing the rest of your on-page setup at the same time, our free meta tag generator handles titles, descriptions, and Open Graph in one pass, and the free SERP preview tool shows how your snippet renders before you publish. For the bigger picture on which structured data still earns features, our guide to a schema markup generator for rich results covers the types that survived 2026.
The Bottom Line#
FAQ schema is not dead. The FAQ rich result is. If your only reason for using FAQPage markup was the SERP accordion, that reason is gone, and you can stop chasing it. If your reason is being understood and cited by AI Overviews, ChatGPT Search, and Perplexity, FAQ schema is more valuable in 2026 than it was before.
Write honest questions, answer them directly on the page, mark them up with valid FAQPage JSON-LD, validate, and ship. The free schema markup generator does the syntax so you can focus on the answers, which is the part that actually moves AI visibility.
Frequently Asked Questions#
Is FAQ schema dead in 2026? No. Google deprecated the FAQ rich result (the expandable accordion in search) on May 7, 2026, so you no longer get extra SERP space from it. The FAQPage markup itself still works and is now a primary input for AI Overviews, ChatGPT Search, and Perplexity, which parse it to build cited answers.
Does FAQ schema still help SEO? Yes, but indirectly. It no longer earns a rich result, so it will not change how your blue link looks. It does help answer engines find and quote clean question-answer pairs from your page, which improves your odds of appearing and being cited in AI-generated answers.
Will adding FAQPage markup get me penalized? Not if you use it honestly. The rich result was deprecated, not penalized. Google still requires that the questions and answers in your markup are visible on the page and genuinely answer the question. Invented or hidden FAQ markup can still draw a manual action.
How many questions should FAQ schema have? Four to eight genuine questions is a healthy range. Use questions real people ask, pulled from "People also ask," autocomplete, and your support inbox, and keep each answer self-contained in roughly 40 to 60 words so an answer engine can quote it on its own.
What is the difference between the FAQ rich result and FAQPage schema? The rich result was the visual feature Google rendered in the SERP, the expandable list of questions under your listing. FAQPage schema is the structured-data code in your page source. Google removed the visual feature in 2026 but the code remains valid and machine-readable, which is why it still matters.
How do I add FAQ schema to my page? Write four to eight real questions with direct answers, mark them up as FAQPage JSON-LD, and place the script in your page. The simplest path is to paste your pairs into the free schema markup generator, copy the validated JSON-LD it outputs, and add it to your page head, then confirm it parses with Google's Rich Results Test.
