A content team ships a new FAQ section, wraps it in FAQPage schema, and waits for the rich result that used to show up in Google Search. It never comes. Google restricted FAQ rich results to a narrow set of government and health sites back in August 2023, and by May 2026 it had deprecated the feature entirely, pulling the FAQ rich result report and Rich Results Test support along with it.1 Most B2B marketers still building FAQ blocks for the snippet that disappeared three years ago are optimizing for a SERP feature that no longer exists on their site, whatever their site is.
The instinct to redirect that effort toward AI citations instead makes sense, except a new 2026 study just complicated that plan too. Ahrefs tracked 1,885 pages that added JSON-LD schema between August 2025 and March 2026, matched them against 4,000 control pages, and found adding schema produced no meaningful citation lift on any platform: Google AI Mode moved 2.4 percent and ChatGPT 2.2 percent, both statistically indistinguishable from zero, while Google AI Overview citations actually declined 4.6 percent relative to controls.2 Schema markup, on its own, is not the lever a lot of AEO advice claims it is.
None of that means FAQ content is dead weight. It means the value moved from the markup to the writing underneath it. This piece is about what to build instead: an original framework for structuring FAQ-style content so it earns AI Overview and answer-engine citations on its own merits, plus the narrow, specific case where the schema itself still matters.
What FAQ schema was built for, and why the rich result is gone
FAQPage structured data was designed to do one job: tell Google that a page contains a list of questions with single, direct answers, so Google could render an expandable Q&A block directly in the search results.3 For a few years it worked as a real SEO lever, doubling a listing’s visual footprint on the results page for almost no engineering cost. That is exactly why it got abused, with marketing teams stuffing promotional and duplicate FAQ blocks onto pages that had no real reason to carry them, which is also exactly why Google pulled it back.
The restriction in August 2023 limited FAQ rich results to well-known, authoritative government and health websites.1 For every other site, including every B2B SaaS, fintech, cybersecurity, manufacturing, and e-commerce company reading this, the rich result was already gone two and a half years before Google made it official. The May 2026 deprecation notice on Google’s own developer documentation just closed the file: FAQPage is still a valid schema.org type, and unused structured data does not hurt a page, but it will not produce a visible result in Google Search for a general commercial site again.1
The new problem: schema alone does not move AI citations either
The natural next move was to keep the schema and repurpose the pitch: FAQPage markup as an AI citation play instead of a rich-result play. The Ahrefs study above is the most rigorous public test of that idea to date, and it is worth sitting with the caveat as much as the headline number. Every page in the 1,885-page dataset already had more than 100 AI Overview citations before schema was added, meaning the test measured whether schema helps a page that AI systems already trust get cited more, not whether it helps an invisible page get discovered at all.2 For a page nobody is citing yet, schema might still help a crawler parse and index it correctly. It just will not manufacture a citation out of content that was not extractable in the first place.
That distinction matters because it points to where the real leverage sits. Search Engine Land’s own analysis of FAQ schema’s decline reaches the same conclusion from a different angle: the tactic’s surviving value is not the markup, it is the discipline of writing short, structured, factual answers that both large language models and human readers can lift cleanly, whether or not a script tag ever touches the page.3 The schema was never the mechanism. It was a wrapper around content that, done well, was already extraction-ready.
The Visible-Answer Model: a 5-layer framework for FAQ content built for citation, not rich results
The framework below is what we run on client FAQ sections now that the schema itself is no longer the point. Each layer is a build step, run in order, on every question block you want considered for an AI Overview, a People Also Ask expansion, or a direct LLM citation.
Layer 4 is the one teams get backwards. A lot of legacy FAQ schema was written for the rich result, not the reader, which meant the JSON-LD text was often more polished or more compressed than what actually appeared on the page. That gap used to be a minor inconsistency. Under Google’s current structured data guidelines it is treated as a real content violation, since FAQPage markup is only valid when it reflects content that is actually visible on the page, not a separate, schema-only version of the answer.4 If you are keeping FAQPage schema at all in 2026, and the LLMO discipline still has good reasons to, that parity is not optional.
FAQ schema for rich results vs. FAQ content structured for AI extraction
These two goals get treated as the same project because they both start with a bulleted list of questions. The mechanics underneath are different enough that building for one without the other explains most of the wasted FAQ work we see in content audits.
| Dimension | FAQ schema for rich results (2018 to 2023) | FAQ content for AI answer extraction (2024 to 2026) |
|---|---|---|
| Primary goal | Win extra vertical space in the SERP with an expandable Q&A block | Get the exact sentence lifted into an AI Overview, a People Also Ask answer, or a chatbot response |
| What gets evaluated | The JSON-LD markup structure and eligibility of the domain | The visible passage itself: length, self-containment, and whether it stands alone outside the page |
| Required format | Valid FAQPage JSON-LD with question and acceptedAnswer pairs | A question-shaped heading followed immediately by a 40 to 75 word visible answer, schema optional |
| Eligibility | Restricted to well-known government and health sites since August 2023, fully deprecated May 2026 | Open to any page an answer engine chooses to crawl and trust |
| Still works in 2026? | No, the rich result itself no longer renders for a general commercial site | Yes, this is the actual mechanism behind AI Overview and LLM citation today |
Read the last row carefully. Almost every AEO checklist worth following, including our own AEO checklist for featured snippets and AI Overviews, converges on the same right-hand column: question-shaped headings, short self-contained answers, and evidence placed inside the answer block itself. FAQPage schema was never a substitute for that work. It was, at best, a formatting hint layered on top of it.
Where teams get this wrong
The most common mistake is chasing the schema as a fix, adding or re-validating FAQPage markup on pages where the underlying answer is buried in a long paragraph, hedged with three qualifiers, or written in a voice that assumes the reader already read the two paragraphs above it. No amount of correctly formatted JSON-LD repairs a passage that fails the Answer Lock layer. Fix the visible sentence first; the markup, if you use it at all, should be the last five minutes of the job, not the first.
The second mistake is schema-content drift: a JSON-LD block that answers the question more crisply, or completely differently, than the visible text on the page. This usually happens when a developer or an SEO tool generates the schema separately from whoever wrote the copy, and the two never get reconciled. Google’s structured data guidelines are explicit that this is a policy violation, not a stylistic gap, so it is worth an actual audit pass rather than an assumption that the schema is fine because it validates.4 If your team is running any kind of technical and content SEO program, a parity check between visible FAQ text and its JSON-LD twin belongs in the standard release checklist, not a one-time cleanup.
Frequently Asked Questions
What is FAQ schema?
FAQ schema, or FAQPage structured data, is a schema.org markup format that tells search engines a page contains a list of questions with single, direct answers. It was originally built to trigger an expandable Q&A rich result directly in Google Search.
Is FAQ schema markup still worth using in 2026?
The rich result it once produced is gone for general commercial sites, restricted to government and health domains since August 2023 and fully deprecated by Google in May 2026. FAQPage remains a valid schema.org type and can still help crawlers parse a page’s Q&A structure, but it will not by itself win a visible search result or force an AI citation.
Does schema markup help AI citations?
A 2026 Ahrefs study tracking 1,885 pages that added JSON-LD schema found no statistically meaningful citation lift on Google AI Overviews, Google AI Mode, or ChatGPT, and a small decline in AI Overview citations relative to control pages. Schema may still help an undiscovered page get crawled and indexed correctly, but it does not manufacture a citation from content that is not already structured for extraction.
What is answer engine optimization (AEO)?
AEO is the discipline of structuring content so it can be selected as the direct answer to a specific question, whether that shows up as a featured snippet, a People Also Ask expansion, or a Google AI Overview citation, instead of only ranking as a link a searcher has to click through.
How should B2B SaaS platforms structure content for AEO?
Use real buyer phrasing as question-shaped headings, put a self-contained 40 to 75 word answer immediately underneath each one, attach one cited fact or spec inside that same block, and keep any FAQPage schema in exact parity with the visible text. This is the Visible-Answer Model covered above, and it holds regardless of whether the schema renders anywhere.
How do I add FAQ schema in JSON-LD?
Wrap each question and answer pair in a FAQPage object with mainEntity entries of type Question, each containing an acceptedAnswer of type Answer. The text inside acceptedAnswer has to match the visible answer on the page exactly, since Google’s guidelines require the markup to reflect content users can actually see, not a separate or optimized version of it.
Want a page-by-page read on which of your FAQ sections already pass the Answer Lock and Structural Mirror tests, and which ones are quietly failing on schema-content drift? Start with a GEO audit to see where your content stands today.
Sources: [1] Google Search Central, “Mark Up FAQs with Structured Data”; see also Search Engine Journal, “Google Drops FAQ Rich Results From Search,” August 2023. [2] Ahrefs, “We Tracked 1,885 Pages Adding Schema. AI Citations Barely Moved,” 2026. [3] Search Engine Land, “The rise and fall of FAQ schema, and what it means for SEO today,” 2026. [4] Google Search Central, FAQPage structured data general guidelines.
Share this article
Alex Carter leads SEO and content strategy at MV3 Marketing, specializing in generative engine optimization, technical SEO, and AI-driven content systems for B2B companies.
Ready to audit your organic growth opportunity?
$2,500 flat. 5 business days. Six deliverables tied to pipeline , not rankings. No retainer required.
Get the Organic Growth Audit →