Google killed FAQ rich results. Good.

On May 7, FAQ rich results stopped appearing in Google Search. The Search Console reporting goes in June. The API support goes in August. Google didn't publish a blog post. They added a deprecation notice to the documentation and moved on.

The SEO industry's reaction has been a mix of mild surprise and mild grief, which is the wrong reaction. FAQ rich results have been functionally dead since 2023, when Google restricted them to "well-known, authoritative government and health websites." What died this week is the last 3% of the feature. The remaining 97% died three years ago and nobody noticed because the schema kept getting recommended in GEO advice anyway.

That's the actual story. Not the deprecation. The fact that an industry kept telling clients to implement schema for a feature that didn't exist for them, and then quietly pivoted to claiming the same schema helps with AI search instead — without evidence, without testing, and without acknowledging the pivot.

The schema-as-magic mental model

There's a persistent belief in SEO that structured data is a kind of incantation. You add the markup, the search engine reads it, good things happen. The mechanism is rarely specified beyond "machine readability."

FAQ schema sat at the centre of this mental model for years. It was easy to implement. It produced visible results in SERPs. It made pages look bigger and more authoritative on the page. Clients understood it. Agencies billed for it. Tools generated it. Everyone was happy.

When Google restricted FAQ rich results in 2023, the honest response would have been: this is over, stop selling it, find something else. Most of the industry did neither. The schema kept getting recommended. New justifications appeared. *It still helps Google understand your content. It might help with AI search. It's good practice.* None of these were tested. None had evidence behind them. They were post-hoc rationalisations for a tactic that had lost its actual purpose.

This week's deprecation removes the last sliver of plausible deniability. FAQ rich results are gone. The schema can stay on the page — Google has confirmed it won't cause problems — but it no longer does anything visible in Search.

The GEO pivot was always suspect

Watch what happens next. A non-trivial chunk of the GEO industry will claim FAQ schema still matters because *AI systems use it for retrieval*. This claim will be made confidently. It will be made by people selling GEO services. It will not be accompanied by data.

Here's what we actually know. Pedro Dias laid it out plainly a few weeks ago, and the architecture argument doesn't get less true when it's inconvenient: in a probabilistic retrieval system, schema doesn't *ensure* anything. It might help. It might not. The evidence for FAQ schema specifically helping AI citations is, charitably, thin.

What we do have evidence for is the boring stuff. Domain authority. Quality backlinks. Brand recognition. Clean technical delivery. Mike King's recent work on the 499 status code is a useful reminder of where the real eligibility gates sit — pages that time out for AI crawlers get cited 18x less often, which is a much bigger effect than any plausible schema lift. The ground-floor question for AI citation isn't *did you mark this up as an FAQ.* It's *can the agent fetch your page before its budget runs out.*

The honest version of the GEO playbook has always been the SEO playbook plus structured data plus earned media. FAQ schema was never load-bearing in that stack. It was a SERP feature that briefly worked, then didn't, and the industry never updated.

Why Google actually killed it

Google didn't explain. They never do. But you can read the trajectory.

FAQ rich results were a transitional feature. They existed when Google still believed publishers and search were collaborating on the same page. The publisher provides the FAQ markup; Google displays it in a richer SERP unit; clicks happen; everyone wins. That bargain has been unwinding for three years. AI Overviews replaced the answer-extraction job that FAQ rich results used to do. The data is in the 38% click lift when AI Overviews were removed in a randomised test, and in the 60% search-referral collapse Chartbeat measured for small publishers.

If you're Google, why keep a SERP feature that extracts publisher content into a richer card when you've built a system that extracts publisher content into a generated answer? The card and the AI Overview are doing the same job, and the AI Overview doesn't need the publisher to mark anything up. It just reads the page.

FAQ rich results were a feature from a previous era of search. The era ended. The feature was the last thing standing from that era, and now it's gone.

What this means for the work

If you have FAQ schema on your pages, leave it. Google has confirmed it doesn't cause problems. Removing it is busywork. The schema is still a valid Schema.org type. If an AI system ever does use it for retrieval — which I'd put low odds on, but it's not impossible — you'll be glad it's still there. The downside of leaving it is zero. The downside of removing it is wasted dev time.

If a GEO consultant is currently pitching FAQ schema as an AI search tactic, ask them for evidence. Not theory. Not "AI systems probably read structured data." Actual test data showing FAQ markup correlated with citation lift, controlled for the obvious confounds (the pages with FAQ schema also tend to be pages from sites with more authority, which is what's actually doing the work).

If you're building new pages, the question of whether to add FAQ markup is now a pure structural question. Does the page genuinely have FAQs that benefit from being marked up as discrete Q&A units? Add it. Are you adding it because someone told you it helps with AI? Don't bother. The opportunity cost of that engineering time is better spent on Core Web Vitals, on fixing 499s, on the things that actually gate AI eligibility.

The deeper lesson is the one the industry keeps refusing to learn. Tactics don't outlive the systems they were designed for. FAQ rich results were designed for SERP enhancement. SERP enhancement has been replaced by generated answers. The tactic is over. Pretending it has a second life in AI search because the acronym changed isn't strategy. It's inventory clearance.

The schema conversation needs to grow up

FAQ schema dying is a small story. The bigger story is that the industry's relationship with structured data has been broken for years, and FAQ was just the most visible example.

Schema doesn't make pages rank. Schema doesn't make AI systems cite you. Schema does help certain systems parse certain things in certain contexts, and when it works, it works as one input among many — not as a magic enabler. The honest framing is that structured data is part of being machine-readable, which is part of being eligible, which is part of being considered, which is upstream of being cited or ranked.

Most schema advice you see online skips all of that and goes straight to "add this markup to get these results." It's the same shape of advice that powered FAQ schema's persistence three years after the feature was restricted. The advice doesn't update because nobody who sold it has an incentive to retract.

That's the loop. And we built it.

Google killing FAQ rich results is a small mercy. It removes one more piece of zombie advice from the playbook. The next ten pieces are still there, waiting for Google or OpenAI or Anthropic to quietly deprecate the assumptions they were built on. The industry should stop waiting and start auditing the playbook itself.