In May 2026 Google quietly ended FAQ rich results, the expandable questions that used to appear under a search listing. A lot of sites added FAQPage markup purely to win that extra space, so the obvious question is whether to rip it out. The second, more interesting question is whether FAQ content still matters now that answers are increasingly written by AI. My short view: keep the markup if it is accurate, stop expecting it to do much on its own, and put the effort into the visible questions and answers themselves. Every article on my site, including this one, opens with a short answer and closes with FAQs that are also published as FAQPage structured data, so I will explain why I still do that and what I do not expect from it.
What Google actually changed, and when
Google's Search Central documentation changelog has two entries that matter. On 8 May 2026 it added a deprecation notice to the FAQ rich result documentation, stating that the feature would no longer appear in Google Search starting 7 May 2026. On 15 June 2026 it removed the FAQ rich result documentation altogether, noting that the feature is no longer shown in Google Search results. This finished a process that started in 2023, when Google limited FAQ rich results to well-known, authoritative government and health websites, which is why most commercial sites had already stopped seeing them. Secondary coverage, including TechWyse's write-up of the change, reported that the FAQ search appearance, the rich result report in Search Console and Rich Results Test support would be dropped over the summer, with Search Console API support following later. I have not seen those later steps confirmed in Google's own changelog, so check your Search Console account rather than relying on the reported dates. Google did not publish a blog post or a reason alongside the notice.
Is FAQPage markup still worth keeping?
FAQPage is a schema.org type, not a Google invention, and Google removing a search feature does not make the markup invalid. Valid structured data that accurately reflects visible content does not hurt you, and other systems can still read it: Bing, other search engines, and any crawler that parses JSON-LD to understand a page. Google's own guide on generative AI features in Search, published in May 2026, is careful here. It says structured data is not required for generative AI search and that there is no special schema.org markup you need to add, while still recommending structured data as part of a normal SEO strategy. My rule for clients is simple. If FAQPage markup is already in place and matches FAQs that are visible on the page, leave it. It costs nothing to keep. If it marks up questions that are hidden, padded or written only for the markup, remove it along with the hidden content. And do not start a project to add FAQPage markup across a site in the hope of AI visibility, because the evidence does not support that spend.
What the evidence says about schema and AI citations
The most useful study I have found is a controlled one. Ahrefs (Louise Linehan, May 2026) tracked 1,885 pages that added JSON-LD schema between August 2025 and March 2026 and compared them with matched control pages, measuring citations in Google AI Overviews, AI Mode and ChatGPT. Its headline finding was that adding schema produced no major uplift in citations on any platform. AI Mode and ChatGPT moved by roughly 2%, which the authors say is indistinguishable from zero, and AI Overviews fell slightly. Two limits are worth knowing. The pages already had AI Overview citations before the change, so the study says little about pages that are not yet in contention. And schema types were pooled, so FAQPage was not tested on its own. Microsoft has said something more positive about its own stack: Search Engine Land reported that Fabrice Canel of Bing said at SMX Munich in March 2025 that schema markup helps Microsoft's LLMs understand content. Understanding is not the same as citation, though, and no vendor has said schema guarantees a mention. I read the evidence as: markup is a weak signal at best, and content does the heavy lifting.
Why clear question-and-answer content still helps
The case for FAQs was never really about the markup. Assistants answer questions, and when they search the web to do it they pull passages that answer the question being asked. A page that states a question in plain words and answers it directly in the next two or three sentences is easy to lift and easy to attribute. A page that buries the answer in paragraph nine is not. Google's generative AI guide recommends headings that give content a clear structure and emphasises distinctive, non-commodity content, and neither of those depends on markup. On my own site this shows up in two places. Every article opens with a short answer of 40 to 60 words that answers the title directly, and closes with three or four FAQs that handle the follow-up questions people actually ask. The same text appears in my llms-full.txt and in the page's structured data, so a crawler reading HTML, JSON-LD or a plain text file gets the same answer. I cannot tell you that this causes citations. I can tell you it makes the content easier for both people and assistants to use, which is reason enough.
How to write FAQs that are worth citing
Most FAQ sections fail because they are written for keywords, not people. A few rules I follow. Use questions you have actually been asked, from sales calls, support tickets or your CRM notes, in the words the buyer used. Answer in the first sentence, then add the condition or exception. Include something specific: a number from a named source, a threshold, a step, a constraint, because specific claims are what an assistant can quote and generic advice is what it already knows. Keep each answer self-contained, so it still makes sense if it is the only thing someone reads. Do not repeat the article body; FAQs should handle the follow-up questions, objections and edge cases. Keep them visible on the page, not hidden behind tabs that load later. And date anything that will change. When a fact moves, such as a platform feature or a policy, update the answer and the page date together, and say what changed. An honest updated date is a much stronger signal than a page that claims to be fresh and is not.
A simple checklist for existing FAQ markup
If you have a site with FAQPage markup from the rich result era, here is the audit I run. First, crawl the site and list every page with FAQPage JSON-LD. Second, check that each marked-up question is visible on the page. Anything hidden, injected only into the markup, or templated across hundreds of pages with the same answers should go. Third, check the questions themselves. Many FAQ blocks were written to win the rich result and are thin: one-line answers to questions nobody asks. Rewrite the useful ones and delete the rest. Fourth, look at where the most important answers sit. If the answer to the main question of the page lives only in the FAQ at the bottom, move a short version to the top as well. Fifth, stop tracking FAQ rich result impressions as a KPI, because that report is reported to be gone, and replace it with measures of AI visibility you can actually observe, such as visits from assistants and live-answer fetches in your logs. My pieces on tracking AI crawlers and measuring AI search traffic cover both.
Sources
Google Search Central, Latest documentation updates (entries of 8 May 2026 and 15 June 2026): https://developers.google.com/search/updates. Google Search Central, Optimizing your website for generative AI features on Google Search: https://developers.google.com/search/docs/fundamentals/ai-optimization-guide. TechWyse, Google ends FAQ rich results for all sites: https://www.techwyse.com/news/ai-search/google-faq-rich-results-deprecated-2026. Ahrefs, Louise Linehan, We tracked 1,885 pages adding schema. AI citations barely moved (11 May 2026): https://ahrefs.com/blog/schema-ai-citations/. Search Engine Land, Microsoft Bing/Copilot use schema for its LLMs: https://searchengineland.com/microsoft-bing-copilot-use-schema-for-its-llms-453455
FAQ
Yes. Google's documentation changelog says FAQ rich results stopped appearing in Google Search from 7 May 2026, and the FAQ rich result documentation was removed on 15 June 2026.
Not if it accurately marks up FAQs that are visible on the page. The markup is still valid and other systems can read it. Remove it where it marks up hidden, duplicated or thin questions.
There is little evidence that it does on its own. Ahrefs' controlled study in May 2026 found no major citation uplift from adding schema on AI Overviews, AI Mode or ChatGPT, though FAQPage was not tested separately.