TL;DR: As of May 7, 2026, Google stopped showing FAQ rich results in search, and it’s phasing out the related Search Console reporting through the summer. That doesn’t mean FAQ schema stopped mattering. AI systems like ChatGPT, Perplexity, and Google’s own AI Overviews still rely on well-structured Q&A content to decide what gets cited. The move right now isn’t deleting your FAQ markup. It’s auditing it, because most sites have never actually checked whether their still matches what’s on the page.
Google’s May 7, 2026 removal ended a ten-year-old SERP feature, not the schema behind it
FAQ rich results, the expandable question-and-answer dropdowns that used to stretch a listing down the search results page, stopped appearing across all site types on May 7, 2026. Search Console’s FAQ reporting and Rich Results Test support are being phased out through the summer, with API support ending in August.
This wasn’t sudden. Google had already restricted FAQ rich results heavily since 2023, narrowing eligibility down to a small set of government and health sites before pulling the plug entirely this year. The reason is straightforward. The feature got abused. Marketers crammed keyword-stuffed questions into FAQ sections, duplicated the same blocks across dozens of pages, and used the markup purely to grab extra real estate on the results page instead of actually answering anything. Google cleaned house.
A lot of site owners read the announcement and did the obvious thing: ripped the schema out. That’s the part worth slowing down on.
FAQ schema didn’t lose its job. It lost its billboard
Google has been explicit that it will keep using FAQ structured data to understand pages, even without the visual dropdown to show for it. That distinction matters more than it sounds like it should.
AI answer engines work differently than a search results page. When ChatGPT, Perplexity, or Gemini assembles an answer, it’s pulling from the open web and looking for content it can extract cleanly and cite with confidence. A page with a clear question as a header followed by a direct, self-contained answer is about as easy as that gets. FAQ-formatted content maps almost exactly onto how these systems build their own responses.
Here’s the honest, slightly messier part. Independent testing has shown that large language models tokenize the JSON-LD behind FAQ schema as plain text. They don’t parse it the way Google’s crawler does, and they don’t validate whether it’s accurate. One well-known test in early 2026 embedded a fake address inside invalid, made-up schema with no matching visible content, and both ChatGPT and Perplexity extracted it anyway. So schema markup by itself isn’t a guaranteed citation lever. What’s reliably working is the visible, plainly written Q&A content the schema is supposed to describe. Schema adds a second, supporting layer on top of that by telling any system parsing the page exactly which sentence is the question and which one is the answer, removing the guesswork. Skip the visible content and lean on schema alone, and you’re optimizing the part that matters less.
Three FAQ problems hiding in plain sight, until someone actually audits the code
Three patterns show up over and over once you actually go looking:
Dead or mismatched schema. The page got edited at some point and the FAQPage markup didn’t get updated with it, so the schema now describes questions and answers that don’t exist on the visible page anymore. Both Google’s guidelines and AI systems treat that mismatch as a trust problem, not a minor housekeeping issue.
Generic, keyword-stuffed leftovers. FAQ blocks built years ago purely to chase the old rich result, answering questions nobody actually asks, written for the algorithm instead of the reader.
No real FAQ content at all. Plenty of small business sites never built genuine Q&A content in the first place. The rich result used to be the nudge that got people to bother. With that nudge gone, the content gap just sits there, quietly costing AI citation opportunities nobody’s tracking.
None of these shows up by glancing at a page in a browser. They show up when someone actually checks the markup against the content, page by page.
What this means for your business
This is exactly the kind of issue a technical SEO audit exists to catch, and it’s a different muscle than writing good content. We just finished a full audit remediation sprint on our own site here at Austin Code Monkey, working through six audit documents covering schema, performance, and on-page issues. One of the bugs we caught was a noindex flag silently misfiring because of a substring-matching error in the code, the kind of thing that only turns up when someone actually reads through the implementation instead of running a generic scanner.
We’d already learned this lesson once before. When Google pulled native Q&A functionality from Google Business Profile in late 2025, we rebuilt that strategy around FAQPage schema instead of letting the content disappear with the feature. Treating structured data as real infrastructure, something that gets audited and maintained, rather than a one-time decoration, is the difference between a site that keeps getting cited and one that quietly stops showing up.
FAQ
Should I remove FAQ schema from my site now that rich results are gone? No, not automatically. Keep any FAQPage schema that still matches your visible Q&A content. Only remove or fix markup that no longer reflects what’s actually on the page.
Do AI tools like ChatGPT actually read FAQPage schema? They tokenize the whole page either way, schema included, but they don’t parse or validate it as structured data the way Google’s crawler does. The schema is a secondary signal layered on top of clearly written, visible Q&A content, which is what these systems extract most reliably.
What’s the most common FAQ schema mistake found in audits? A mismatch between the schema and the visible page content, usually left over after a page has been edited and nobody has updated the markup to match.
How do I check if my FAQ schema is helping or hurting? Run the page through Google’s Rich Results Test, then compare the marked-up questions and answers line by line against what’s actually visible on the page. Any mismatch needs fixing right away.
Is this a content problem or a technical problem? Both. Writing genuine, specific Q&A content is the content half. Finding every page where the schema doesn’t match that content anymore, across an entire site, is the technical half, and that part needs an actual audit since it’s invisible otherwise.
Austin Monk, “El Jefe Of SEO,” is a Search Engine Expert at Austin Code Monkey, an SEO and AI search optimization agency in Austin, Texas. Austin has 18+ years of SEO experience and has been ranking businesses in Austin and throughout Central Texas, driving organic growth for service businesses.
