May has been a busy month for anyone watching how Google is reshaping search. Two announcements landed within a week of each other, and together they say more about where Google is going than either one does on its own. The first is the new official guide on optimising for AI Overviews and AI Mode. The second is the quiet retirement of FAQ rich results.
Both are worth reading properly before reacting. Here’s what I took away, and what it changes for the way we approach client work.
FAQ Rich Results Are Officially Done
On 7 May 2026, Google stopped showing FAQ rich results in Search. The expandable question-and-answer dropdowns that used to sit under listings on suitable pages, the ones that bought extra vertical space and a bit of CTR lift, are gone.
The wind-down has a clear timeline. The visible feature went on 7 May. In June 2026, Google retires the FAQ reporting in Search Console and removes FAQ support from the Rich Results Test. In August 2026, the Search Console API stops returning FAQ data altogether.
A few things to clear up, because the headlines have caused some panic:
- FAQPage schema is not banned. It’s still valid markup. You can leave it on your pages without any issue.
- Pages with FAQ schema are not being penalised. Nothing happens to rankings.
- Google has said the data still helps it understand the page. The markup is just no longer tied to a visible SERP feature.
So the practical question for any site currently using FAQ schema is whether to keep it or pull it. There’s a case both ways. If the markup was added mainly to chase the rich result, it’s no longer earning its keep on that front. If the FAQ content on the page is genuinely useful for users and the schema accurately describes it, leaving it in place gives Google a signal that’s confirmed to still be used. Other AI systems and search engines may also parse it for their own purposes.
The AI Search Guide Lands a Few Days Later
On 15 May 2026, Google published its first official guide to optimising for generative AI features in Search. It’s called “Optimizing your website for generative AI features on Google Search”, and it covers AI Overviews, AI Mode, and the agentic experiences emerging alongside them.
The one line everyone is quoting: from Google’s perspective, optimising for generative AI search is just SEO. That’s already being used in some posts to argue that GEO and AEO were never real. I don’t think that’s quite right, and the guide itself is more useful than the takes around it.
Google’s point is simple when you strip it back. AI Overviews and AI Mode are built on top of the regular Search index. Same crawling, same ranking systems, same content. If a page can’t be indexed and shown with a snippet in normal Search, it’s not going to appear in an AI Overview either.
So the foundations matter as much as they ever did. Crawlable site, indexable pages, decent technical performance, content that actually says something. That part isn’t new.
Where Google pushes back is on the idea that you need a separate technical toolkit to be visible in its AI features. The guide names four things that don’t move the needle for Google AI search:
- llms.txt files and other “special” markup. Google might crawl them, but they aren’t a signal for AI Overviews.
- Content chunking. No ideal page length, no need to break content into tiny extractable blocks.
- Rewriting content just for AI. You don’t need every heading as a question or pages built around fan-out variations.
- Inauthentic mentions. Manufactured mentions across low-quality blogs and forums don’t help.
Schema gets a softer mention too. Not required for AI visibility on Google, still useful for rich results where they exist.
The list is a helpful clarification. Plenty of advice has been circulating about llms.txt, chunking, and AI-specific markup as must-haves. Google is saying clearly that for its own AI surfaces, they’re not required. The picture across other AI platforms is more nuanced, which we’ll come back to.
What These Two Updates Say Together
Read on their own, each announcement is a tidy piece of housekeeping. Read together, they tell a clearer story about Google’s direction.
The FAQ retirement is the latest in a longer list of structured data features Google has wound down. HowTo went in September 2023. In June 2025, Google retired seven more (Book Actions, Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement, Vehicle Listing) on the same reasoning, that they weren’t widely used and no longer added meaningful value for users. FAQ is just the most widely used one to go. Read together, the direction looks consistent: markup-based SERP enhancements are quietly being phased out.
At the same time, Google is making AI answers the centrepiece of the results page. AI Overviews now appear on roughly half of all queries. AI Mode is rolling out as a full chat-based alternative to the SERP. The same week Google removed FAQ dropdowns, it published a guide telling site owners that no special AI markup is needed to appear in those AI answers.
The thread connecting both moves is the same. Google is reducing the surface area where webmasters can game presentation through markup, and putting more of the answer experience under its own control. Content quality, topical authority, and how well a page actually satisfies a real question are what’s left to work with.
Where GEO and AEO Still Do Real Work
Here’s the bit the headlines tend to skip. Google’s guidance is about Google’s surfaces. It doesn’t cover ChatGPT, Perplexity, Claude, or Copilot.
Those systems work differently. Different models, different retrieval, in some cases different signals. A brand that wants to be cited in ChatGPT responses or recommended by Perplexity is playing a related game, but not the same game.
That’s where proper GEO and AEO work still matters. Not as a separate discipline bolted onto SEO, but as an extension of the same strategy across every surface where a customer might end up. Same content quality bar. Same technical foundations. Different signals to track, different platforms to monitor, and a wider set of citations and mentions to build.
The Google guide actually makes the case for this kind of work clearer, not weaker. By spelling out what doesn’t work, it makes the things that do work easier to defend: original perspective, first-hand experience, third-party authority, clean technical foundations, strong brand entity. Those are what AI engines reward.
What Changes for Strategy
For most of the clients we work with, very little changes. If the content strategy was built around expert input, real data, and a clear point of view, it was already pointing in the right direction. If the technical SEO was solid, it stays solid.
A few practical things to think about after these two updates:
- Don’t panic about FAQ schema. Leave it where it’s accurate and useful. Remove it where it was added purely for the rich result and isn’t reflecting the actual page.
- Adjust your reporting. If your dashboards track FAQ rich result appearances or use the Search Console API for FAQ data, plan around the June and August retirements.
- Don’t pay for llms.txt subscriptions for Google AI Overviews. It’s not a Google lever. There’s a separate case for one with Claude or smaller AI surfaces.
- Keep investing in original, expert-led content. This is the single biggest lever both updates point toward.
- Look at AI visibility across platforms, not just Google. ChatGPT, Perplexity, and Copilot still reward work that goes beyond standard SEO.
So Where Does That Leave Us
Both announcements point the same way. Markup shortcuts are losing ground. Original content built on real experience and authority is what AI engines, plural, keep rewarding. None of that is new advice. It’s just clearer than it was two weeks ago.
If you’re trying to work out what these changes mean for your site, whether your FAQ schema is still pulling its weight, or where your GEO and AEO investment should sit after the new guide, get in touch and we’ll take a look.
