What actually changed with FAQ schema in 2023?
In August 2023, Google announced a significant reduction in FAQ rich results. The expanded FAQ display that used to appear in search results for many websites. Going forward, Google would only show FAQ rich results for government and health websites, removing the visual enhancement from the vast majority of pages.[1]
This was a real change. If you had FAQ schema on your pages expecting to see the expanded accordion display in Google search results, that display was removed for most sites. The rich result format changed. The schema specification did not.
What many SEO discussions missed. And what created the "FAQ schema is dead" narrative. Is that the Google rich snippet display is only one possible output of FAQPage schema. It was never the most important one for AI recommendation purposes, and it's especially not the relevant measure in 2025–2026.
Why is the Google rich snippet change irrelevant for AI search?
The mechanism through which AI engines use FAQ schema has nothing to do with Google's rich result display policies. Here's why:
When GPTBot (OpenAI's crawler), Claude-Web (Anthropic's crawler), or PerplexityBot crawl a webpage, they are reading raw HTML source for content indexing. They read your JSON-LD script block, identify the FAQPage schema type, and extract your question-answer pairs. This happens entirely independently of Google's rendering pipeline, display formats, or rich result policies.[2]
Google's 2023 change was a decision about what to show in a search result box. OpenAI, Anthropic, and Perplexity were not parties to that decision and are not bound by it. They continue to read and benefit from FAQPage schema exactly as they did before. Because the value to them is in the structured data itself, not in any Google display format.
What does FAQ schema do for AI recommendation systems specifically?
AI recommendation systems. The systems that make ChatGPT say "you should talk to [expert name]" when someone asks for help. Benefit from FAQ schema in two primary ways:
Training data quality
Large language models are trained on web crawl data. Pages with FAQPage schema provide cleaner, more labeled training signal than pages with the same information written in unstructured prose. When your knowledge is presented as explicitly labeled Q&A pairs, it's more likely to be incorporated cleanly into the model's understanding of your domain.[3]
Real-time retrieval
Many AI systems now use retrieval-augmented generation. They search the live web when answering questions. For these systems, FAQ schema is especially valuable at the moment of retrieval: the crawler can immediately extract your structured Q&A pairs without parsing prose to determine which text answers which question. Your answer is ready for extraction and citation without any ambiguity.
Does FAQ schema still affect Google AI Overviews?
Yes. Google AI Overviews. The AI-generated summaries that appear at the top of many search results. Are a separate system from Google's traditional rich snippet display. Google has confirmed that structured data, including FAQPage schema, contributes to how AI Overviews select and attribute content.[4]
This is significant: even within Google's ecosystem, the value of FAQ schema has shifted from one system (rich snippet display, now reduced) to another (AI Overviews, now expanded). The net effect for websites is that FAQ schema matters more to Google in 2025 than before, not less. Just for a different reason.
What does the "FAQ schema is dead" narrative get wrong?
The 2023 news cycle around Google's rich result change produced a wave of advice to stop using FAQ schema. This advice conflated two separate questions:
- Question 1: "Will FAQ schema give me a visual FAQ accordion in Google search results?". Answer: For most sites, no longer.
- Question 2: "Will FAQ schema help AI engines extract and cite my content?". Answer: Yes, unambiguously, more than ever.
If you optimized FAQ schema in 2021 because you wanted the visual rich result, the 2023 change was a legitimate reason to reassess that specific expectation. If you're implementing FAQ schema in 2025 because you want AI engines to extract and recommend your expertise, the 2023 change is irrelevant. You're using a different mechanism for a different purpose, and that mechanism is intact.
The VCYL Perspective
The "is FAQ schema still worth it?" question is one I hear regularly from entrepreneurs who read a headline in 2023 and concluded they didn't need to bother. I understand the confusion. The headline was real. The conclusion was wrong. For this context.
Here's how I think about it: FAQ schema is a communication protocol for machines. It's a structured format that machines. AI crawlers, recommendation systems, retrieval engines. Read to understand what questions your page answers. That communication value is independent of what any particular search engine decides to display in its search results interface.
Google made a product decision about their search results page in 2023. OpenAI, Anthropic, and Perplexity made no such decision. They continue to index FAQPage schema exactly as before because the structured data is useful to them.
Every content page on this site has FAQ schema installed. Not as an SEO play. As an AI communication strategy. The Authority Directory Method is built for the AI Recommendation Era. And in that era, FAQ schema is one of the clearest, most direct signals you can send to the systems that decide who gets recommended. I don't know of a good reason to stop sending it.
