For years, marketers treated FAQ schema as the fast lane to answer visibility. Wrap questions in markup, validate the code, and wait for the accordion to appear in search. That shortcut no longer exists. Google has retired FAQ rich results entirely, and AI assistants like ChatGPT, Gemini, and Perplexity never relied on your markup anyway. Answer selection today is a content quality decision, not a code decision. This blog explains what answer engines actually evaluate when they pick one source over hundreds, and how to restructure your pages so your answers get selected and cited.
On May 7, 2026, Google added a deprecation notice to its structured data documentation confirming that FAQ rich results no longer appear in Google Search, with related Search Console reports and Rich Results Test support removed in the following months. This completed a withdrawal that began in August 2023, when eligibility was restricted to authoritative government and health sites.
The deprecation confirmed what practitioners long suspected: FAQ markup was a display feature, not a selection signal. It never told Google, or any AI system, that your answer was accurate, complete, or worth repeating.
Meanwhile, the stakes have grown. A Pew Research Center analysis of real user browsing data found that when an AI Overview appears, users click a traditional result only 8 percent of the time, against 15 percent when no summary is shown. If the answer box absorbs attention, being the source inside it is the visibility that matters. Markup alone was never going to get you there.
Answer engine optimization, or AEO, is often reduced to a checklist: add FAQPage markup, add Q&A headings, validate, publish. The problem: this checklist describes formatting, while selection is driven by retrieval and trust.
The distinction that matters:
Google’s systems moved to passage-level understanding years ago, so they can rank a specific paragraph inside a long page without markup pointing at it. LLMs go further. When ChatGPT or Perplexity answers with citations, a retrieval layer pulls candidate passages, and the model synthesizes from those it judges most relevant and reliable. We unpack this mechanism in our guide on how LLMs decide which content to show in search answers.
In both cases, the unit of competition is the passage, never the markup wrapped around it.
Across Google AI Overviews, ChatGPT search, Perplexity, and Gemini, the same qualities keep separating cited sources from invisible ones.
Answer engines extract. They do not wait for you to build up to a point. A selectable passage states the answer immediately, then supports it. If your page asks “How long does Salesforce implementation take?” and the answer arrives four paragraphs later, the extractable passage does not exist. A reliable pattern: question as a heading, a 40 to 60 word answer underneath, then depth for readers who want more.
Retrieval systems lift passages out of context. A paragraph that leans on earlier sections (“as mentioned above”) loses meaning when isolated, so it rarely gets used. Strong AEO content reads correctly even when one paragraph is quoted alone. Name the subject explicitly, repeat the entity instead of relying on pronouns, and keep one idea per paragraph.
Vague content is hard to cite because it asserts nothing checkable. Concrete details give machines something to anchor on:
Editorial honesty becomes an asset here. Content that overstates certainty tends to conflict with other sources, and conflict lowers the odds of selection.
LLMs weigh consensus. An answer that aligns with credible sources, while adding original clarity, is safer for a model to repeat than an outlier claim. That check extends to your brand. Consistent naming across your site, profiles, and third-party mentions helps systems connect your content to a trustworthy entity. Structured data still contributes here, as we explain in our breakdown of structured data in AI SEO. Organization, Author, Article, and Product schema help machines resolve who is speaking and why they are credible. That is schema doing real work, unlike FAQ markup chasing a retired display feature.
Retrieval depends on parsing. Logical heading hierarchies, descriptive subheadings, scannable lists, tables for comparative data, and fast server-rendered HTML all raise the probability that your passages are indexed and understood. None of this is exotic. It is disciplined on-page work done with extraction in mind.
This does not mean deleting your FAQ sections. Q&A formatting remains one of the most extraction-friendly structures you can publish, because it mirrors how people phrase queries in AI assistants and voice search. What changed is the reason it works.
Keep the questions and the concise answers. Stop expecting the JSON-LD wrapper to do the selecting. A well-written answer with no markup beats a thin answer with perfect markup, in classic results and AI answers alike. For a broader view, see our primer on what AEO is and why it matters.
Run your key pages through six checks:
Teams that run this audit regularly turn answer visibility into a measurable channel.
AEO does not replace SEO. The pages AI systems cite most often usually earn strong organic visibility too, because crawlability, authority, and depth feed both outcomes. The practical move: build pages that rank in classic search and are structured for extraction into AI answers.
That is a different skill set from markup deployment, and where specialist support pays off. TIS delivers answer engine optimization services covering AI platform query research, passage-level restructuring, entity and schema strategy, and citation tracking across ChatGPT, Gemini, Perplexity, and Google AI Overviews. And because answer selection builds on search fundamentals, partnering with an experienced seo services company keeps the technical and authority groundwork beneath your AEO program sound.
FAQ schema was a formatting instruction. Answer selection is a trust decision. Google’s retirement of FAQ rich results removed the last incentive to confuse the two. The sources winning citations in AI answers share the same traits: direct answers placed first, self-contained passages, verifiable specifics, consistent entity signals, and clean architecture machines can parse. Rebuild your key pages around those traits and you compete for every answer surface at once. Keep optimizing for a retired rich result and you compete for none.
Want to know where your content stands? Request an AI visibility audit from TIS. We will show you which questions you can win, and what it takes to get selected.
It is the practice of structuring content so AI systems like ChatGPT, Gemini, Perplexity, and Google AI Overviews select and cite it when generating answers. SEO targets rankings on results pages; AEO targets inclusion inside the answer itself.
FAQ markup no longer produces any visible feature in Google Search, and it was never a selection signal for AI answers. Existing markup causes no harm, but adding it delivers no benefit. Invest in well-structured Q&A content instead.
Retrieval systems pull candidate passages that closely match the query, then the model favors passages that are self-contained, specific, current, and consistent with other credible sources. Entity signals, such as consistent branding and author credentials, increase citation likelihood.
Lead with a direct answer of roughly 40 to 60 words under a question-style heading, then expand. This gives extraction systems a clean, quotable passage while still serving readers who want depth.
Yes. Organization, Author, Article, Product, and similar entity-focused schema help machines understand who published the content and why it is credible, which influences selection. FAQ markup only controlled a display feature Google has retired.
Query the AI platforms with your priority questions, monitor referral traffic from chatgpt.com and perplexity.ai, and use AI visibility tools to log citations.
© 2026 TIS. All rights reserved.
