Full Definition
FAQPage schema is a structured data type, implemented in JSON-LD, that marks up a webpage's question-and-answer content in a machine-readable format. Each question and its corresponding answer are explicitly labeled, allowing search engines and AI platforms to extract them individually rather than parsing unstructured body copy.
In traditional search, FAQPage schema was used to generate rich results, displaying individual Q&A pairs directly in search results pages. Google removed this feature entirely in May 2026, following an earlier restriction in 2023 that had already limited FAQ rich results to government and health sites only. FAQPage schema no longer produces any visible Google Search feature for any site.
For AI-powered search, its role is more nuanced and less settled than most AEO content suggests. A 2025 study by Ahrefs tracking 1,885 pages found no meaningful citation uplift from adding JSON-LD schema on any AI platform, and direct testing confirmed that ChatGPT, Claude, Perplexity, Gemini, and Google AI Mode all extract only visible HTML content during real-time fetching, treating JSON-LD as plain text rather than as specially structured data. The exception is Google's and Bing's own infrastructure, where dedicated pipelines do process structured data for entity disambiguation and knowledge graph inclusion, separate from how they generate an AI answer. In a 2025 study, Ahrefs analyzed 1,885 pages across five major AI platforms and found zero meaningful citation uplift from JSON-LD schema on any of them. (Ahrefs, 2025)
What does hold up, and where the real value lies, is the visible FAQ content itself. A page with clearly written question-and-answer pairs, each self-contained and directly answering a real buyer question, performs well across every AI platform regardless of whether FAQPage markup is present. The markup may contribute at the corpus level, where FAQ-style content has been harvested at scale to build training datasets, but the visible content structure is the primary driver of citation behavior. Implementing FAQPage schema on a page with strong visible Q&A content is reasonable. Treating the markup itself as the citation lever is not.
FAQPage schema is most worth implementing when the questions reflect actual buyer queries, when each answer is fully self-contained, and when the same content is clearly visible on the page. The schema and the visible content should always match exactly. Adding markup to content that is hidden, collapsed, or only present in the script tag is both a technical error and a waste of implementation effort.
Should I remove FAQPage schema from my site now that Google has removed FAQ rich results?
No. Removing it creates more work for no benefit. Google has confirmed that unused structured data does not cause Search problems, so leaving it in place is harmless. The schema may still contribute at the corpus level where FAQ-style content has been harvested into AI training datasets, and it remains valid markup for Google's and Bing's entity disambiguation pipelines. The case for keeping it is stronger than the case for removing it. What you should stop doing is treating FAQPage schema as a primary visibility tactic or investing significant effort in adding it to pages where it doesn't already exist. The visible Q&A content it marks up is what matters, and that content earns citations with or without the schema wrapper.
If FAQPage schema doesn't improve AI citations directly, why implement it at all?
Two reasons remain valid. First, the visible Q&A content that FAQPage schema marks up is the right format for AEO regardless of the schema: self-contained question-and-answer pairs are exactly what AI retrieval systems extract and cite most reliably. The schema is a byproduct of writing good content, not the reason to write it. Second, FAQPage schema still has a role inside Google's and Bing's own infrastructure, where dedicated pipelines process structured data for entity understanding separate from how they generate answers. For companies that appear in Google AI Overviews or Bing Copilot responses, the schema contributes to how accurately those systems characterize the company. That's a narrower benefit than it used to be, but not zero.
What should FAQPage schema questions be based on?
Real buyer queries, specifically the questions that buyers actually type into AI tools when researching a problem or vendor in your category. The most effective FAQPage implementations map directly to a prompt map: if you know that buyers ask "what is the best prior authorization software for specialty practices," a page that answers that question in a self-contained FAQ block, with FAQPage schema marking it up, is positioned to be extracted and cited in exactly the context where the buyer is actively researching. Questions based on keyword research rather than actual AI query behavior tend to produce FAQ content that ranks for search terms but misses the conversational queries that drive AI citations.