FAQ removed: why your schemas cost money for zero return (Ahrefs 2026 study)
Summarize this article with AI
A client calls me on a Friday morning. He’d bet 4,200 € a year on FAQ schema.
A client runs an e-commerce site with 120,000 pages. For three years, he’s injected 4,200 € annually to maintain FAQ markup across all his product sheets. His hope? To appear in AI Overviews. To gain citations.
He’d read reports from « GEO experts » promising that structured markup would increase the odds of being cited by AI. His historical SEO agency charged him 350 € a month to « maintain and optimize markup. » So he held on.
We ran the crawler. Result: 87% of pages still had the markup. Yet, across the 50,000 queries I track, not a single additional citation came from this markup. Zero. The FAQ schema was dead weight.
We decide on a radical test. Disable the schema. Remove the JSON-LD blocks. Redirect the 4,200 € elsewhere.
Six months later, average organic click-through rate climbed 17%. The AI Overviews? Stable. Zero loss. The client gained time, money, and clicks. Just by stopping believing schema would save him.
You might have this same problem. I see it every week.
FAQ in SERP: what Google removed in March 2026
FAQ SERP visibility began disappearing in 2023. Google first restricted rich display to government and health sites only. Then, in March 2026, the rich result was cut for everyone. It’s not a bug. It’s a product decision.
John Mueller, Search Advocate at Google, announced it on Reddit: « Markup types come and go, but only a few deserve to be kept. »
« Google removed FAQ rich results, HowTo, Course Info, Claim Review… The list grows. Each time, the markup remains valid, but the visual result vanishes. A lesson: pouring massive investment into one schema type for SERP gain is a risky bet. »
FAQ schema had become an SEO reflex. Too common. Too unreliable. Google decided. For you, the e-commerce merchant, this means hundreds of hours spent coding Question and Answer blocks serve no purpose for SERPs anymore. Your page will no longer display a small chevron with a dropdown answer. You’ve lost that. Permanently.
And the Ahrefs study, published four days later, closes the debate on the other justification: AI citations.
1,885 pages tested by Ahrefs: schema doesn’t boost AI citations
Ahrefs ran a controlled test. 1,885 web pages that added JSON-LD. Each paired with a control page without schema. The researchers measured citation change across three environments: Google AI Overviews, Google AI Mode, and ChatGPT.
The results?
- Google AI Overviews: -4.6% citations
- Google AI Mode: +2.4%
- ChatGPT: +2.2%
These shifts are so small they fall within statistical noise. JSON-LD schema produced zero measurable effect on AI citations. Even in AI Mode, where we see a slight gain, the boost is minimal and not statistically significant.
This isn’t theory. This is measurement on real sites. 1,885 pages is a massive sample for an SEO test. Ahrefs’ conclusion is unambiguous: adding schema generates no boost in citations.
The « schema for AI » was a sales promise. The data debunks it. You gain by knowing this now: it saves you from continuing to burn budget.
In fact, the negative impact in AI Overviews (-4.6%) even suggests that markup can sometimes complicate how language models parse your content. By contrast, a well-structured page without schema remains perfectly readable to an LLM.
Why do we keep adding schema?
The reflex persists. Many SEO audits still list FAQ markup as best practice. WooCommerce extensions inject it by default. Agencies bill for it. We don’t question what’s « recommended. »
Yet it’s a trap. Coding schema isn’t free:
- Developer time for integration
- Maintenance when Google changes the rules
- Opportunity cost: each hour on markup isn’t spent on internal linking
In my audits, I see 40% of FAQ schemas contain errors. Code sending contradictory signals to crawlers. Worse than useless—it can backfire.
And worst of all is the anchoring effect. You think you’ve « done technical SEO » because you’ve checked a box. You miss the real levers: architecture, silos, pillar pages, semantic anchors.
FAQ markup is dead. You gain by removing it and using your resources where returns are real.
The DOSE framework: recognizing loss to invest better
I apply the DOSE framework taught by Guillaume Attias (BMO Academy). It starts from a simple principle: every loss hides potential gain, if you reallocate wisely.
The loss here is clear:
- Disappearance of FAQ rich results = less SERP real estate
- No correlation between schema and AI citations = end of a sales pitch
But these losses free a precious resource: budget, technical time, focus.
With this client, the 4,200 € annual redirected to internal linking work. Three months building semantic silos: groups of 30-50 pages linked through exact-match anchors around a buying intent.
Result? +17% organic clicks. No extra spending. Zero loss in AI citations.
Another site, 800,000 pages. 2,300 € annual schema maintenance. We converted that budget into 7 thematic silos. Six months later, Google Discover traffic from those silos jumped 34%.
The gain isn’t magic. It’s mechanical. When you stop feeding a task with zero return, you give power to what actually works.
What to build instead: your semantic architecture
No schema, then. But what?
Google, like LLMs, reads your site as a graph. Links between your pages are the skeleton. Link anchors are the instructions.
- A properly woven semantic silo tells the crawler: « these 30 pages discuss the same topic, and this one is the deepest. »
- A precise anchor (« bioclimatic pergola price ») weighs more than « click here. »
- A pillar page that aggregates sub-pages sends a stronger authority signal than any optimized meta description.
The Product schema with reviews and price keeps its value for classic rich snippets. Keep it when it delivers tangible SERP display (stars, availability). Stop giving it a role it doesn’t have.
The budget freed by dropping FAQ markup: you invest it in a silo script. A freelance developer can, in 15 days, restructure 1,000 pages into thematic silos and generate coherent linking. For the cost of 12 months of useless schema maintenance.
Result: positions climb mechanically, click-through improves, and your structure withstands the next algorithm shift. Because Google will keep valuing how information is organized.
FAQ schema was decoration. Semantic architecture is the frame.
What are you stopping today to win tomorrow?
What if your schema costs more than it earns?
I audit your site in depth. In 45 minutes, you’ll know exactly where to stop losing money and where to invest for real wins.
Book a strategic call — 45 minFrequently Asked Questions
Should I delete all my FAQ schemas?
If you’re not a government or health site, yes. FAQ markup brings zero SERP visibility and doesn’t boost AI citations according to Ahrefs. Removing JSON-LD frees maintenance time. Keep only what’s useful: Product, Breadcrumb, LocalBusiness, etc.
Does Product schema help with AI Overviews?
Ahrefs didn’t test Product specifically. But the broader pattern shows markup doesn’t generate an AI citations boost. Product schema remains useful for classic rich snippets (price, availability, stars). Keep it for that reason, not for AI.
What should I spend FAQ schema budget on instead?
Invest in internal linking. Build semantic silos: groups of pages linked through precise anchors around one buying intent. A client redirected 4,200 € here and gained +17% organic clicks. The effect lasts much longer.
Is the Ahrefs study reliable for e-commerce sites?
It covers 1,885 pages across varied sectors. The sample size and control-group design give strong credibility. Results align with what I observe at my clients: zero correlation between schema and AI citations.
How do I quickly audit my current schema’s impact?
Use Search Console to cross-reference clicks on your FAQ-marked pages before and after Google’s removal. If you never got rich results, that’s a clue. Check AI Overview citations via Semrush or Sistrix. If nothing moves, you can cut without worry.
