Your developer added JSON-LD schema markup to your site three months ago. Your SEO agency told you it would help with AI search. You're still not showing up in ChatGPT or Google AI Overviews. What's going on?
The honest answer — one many vendors won't tell you — is that adding schema markup alone is unlikely to move your AI citation rates. A 2026 Ahrefs study tracking 1,885 pages that added JSON-LD found no meaningful citation lift across ChatGPT, Perplexity, or Google AI Overviews. But the story is more nuanced than "schema doesn't matter." One type of schema consistently outperforms the rest. And schema's real value for B2B companies is foundational, not promotional.
Here's what the data says — and what it means for how you should actually allocate your team's effort.
In 2026, Ahrefs tracked 1,885 web pages before and after they added JSON-LD schema markup, matching them against 4,000 control pages. The researchers measured citation changes across Google AI Overviews, Google AI Mode, and ChatGPT over a multi-month window.
Adding schema didn't help. On AI Overviews, it produced a marginal negative result. On ChatGPT and other platforms, the effect was effectively zero.
This contradicts what you'll hear from many SEO tools and agencies. It doesn't mean schema is worthless. It means the causal story most vendors tell — "add structured data, get cited" — isn't supported by controlled data.
Here's where it gets counterintuitive. Pages that are cited by AI systems are roughly three times more likely to have JSON-LD markup than pages that aren't cited. If schema doesn't drive citations, why the gap?
The answer is correlation, not causation. Technically sophisticated organizations — the ones that maintain proper schema markup, keep their structured data current, and invest in site infrastructure — also tend to produce better content, earn more authoritative backlinks, and do everything else that drives AI citations. Schema is a marker of a well-maintained site. It's not the cause of the citations those sites earn.
"AI-cited pages are 3x more likely to have schema. But pages that added schema saw no citation lift. The difference: one is a trait of quality sites; the other is a causal test. Only one of those tells you what to do."
The Ahrefs researchers also tested whether major AI systems actually read JSON-LD during live page retrieval. They tested five platforms: ChatGPT, Claude, Perplexity, Gemini, and Google AI Mode. Every system extracted only visible HTML content. JSON-LD in the page head was ignored during real-time fetching.
There's a meaningful exception to the "schema doesn't directly drive citations" finding, and it matters for B2B companies: FAQPage schema.
FAQPage schema structures question-and-answer content in a format that maps directly to how LLMs extract and paraphrase content. When a page has genuine Q&A content marked up with FAQPage schema, AI systems can identify discrete answer units — complete, standalone responses to specific questions — and quote them with higher precision.
Research into LLM citation patterns consistently shows FAQ-structured content outperforming prose for citation across verticals. This makes intuitive sense: an LLM trying to answer "what is the best approach to X" has an easier extraction job from a clearly labeled Q&A block than from flowing paragraphs that happen to contain the answer somewhere.
The implication for B2B companies: if you're going to prioritize one structured data investment for AI search, FAQPage schema on pages with genuine question-and-answer content is the highest-leverage choice.
Organization schema doesn't directly boost your AI citation rate either. But it performs a different function that matters enormously for B2B brands: entity disambiguation.
AI systems are trained on internet-scale data and hold internal representations of millions of companies. Without clear entity signals, an LLM answering a question about your industry may conflate your company with a competitor, cite an outdated description of what you do, or omit you entirely because it can't confidently identify who you are.
Organization schema tells AI systems — and the search engines that feed them — exactly who you are, what you do, where you operate, and how to find you. It doesn't make you more likely to be cited; it makes you more likely to be cited correctly. For a B2B company spending money on GEO, ensuring your entity is cleanly defined in structured data is foundational to everything else you build.
Microsoft confirmed in March 2025 that schema markup helps their LLMs understand content structure for Bing Copilot responses. Google's Search team stated in April 2025 that structured data provides an advantage in understanding and representing content. The channel is indirect — through indexing and model training, not live retrieval — but the signal is real.
Given the evidence, here's how B2B marketing leaders should think about schema investment:
| Schema Type | Priority | Why It Matters |
|---|---|---|
| Organization | Do first | Defines your entity unambiguously. Prevents AI misidentification. Foundational for everything else. |
| FAQPage | Do second | The only schema type with consistent evidence of direct AI citation benefit. Apply to any page with real Q&A content. |
| Service | Do third | Defines what you offer in structured form. Helps AI systems match your services to relevant buyer queries. |
| BlogPosting / Article | Do fourth | Signals content type and authorship. Good SEO hygiene; indirect AI benefit through indexing quality. |
The Ahrefs schema study is a useful corrective — it redirects effort away from a low-leverage tactic and toward what actually moves citation rates.
The Princeton KDD 2024 GEO research paper (Aggarwal et al.) tested 10,000 queries to find what consistently increased AI citation rates. The findings are clear: statistics with source attribution produced a 41% citation lift and expert quotes with credentials produced a 28% lift. Structural changes — adding FAQs, using direct-answer formats, establishing entity clarity in the opening paragraph — produced consistent improvements across all platforms tested.
Schema supports this work. It doesn't replace it. A page with FAQPage schema and no real Q&A content won't get cited. A page with excellent Q&A content, attributed statistics, and verifiable expert quotes will outperform whether or not the schema is present — though adding FAQPage schema to that content will improve extraction accuracy.
Schema markup is the plumbing of AI search visibility, not the fuel. You need it in place. It needs to be correct. And FAQPage schema deserves priority attention because it's the one type with direct citation evidence behind it.
But if your team is choosing between "spend two weeks perfecting our JSON-LD" and "spend two weeks restructuring our top 10 landing pages with direct answers, attributed stats, and FAQ sections" — the content work wins every time. Get the plumbing right, then focus your optimization energy where the data says it matters.
Adding JSON-LD schema to a page does not directly and reliably increase AI citation rates on its own. A 2026 Ahrefs study tracking 1,885 pages that added schema found no meaningful citation lift in ChatGPT, Perplexity, or Google AI Overviews. The exception is FAQPage schema, which maps directly to how LLMs extract Q&A content and consistently outperforms other schema types for AI citation. Schema is best understood as foundational infrastructure — it supports everything above it but is not itself the citation trigger.
FAQPage schema is the most consistently impactful schema type for AI search citation. It structures question-and-answer content in a machine-readable format that AI systems extract and paraphrase directly. Organization schema is the second priority: it helps AI systems unambiguously identify who your company is, what it does, and who it serves — which prevents misidentification and ensures your brand entity is correctly represented across AI-generated answers.
When AI systems fetch a page during live retrieval, research has found that ChatGPT, Claude, Perplexity, Gemini, and Google AI Mode all extract only visible HTML content — not JSON-LD markup in the page head. Schema's role is indirect: it improves how Google's crawlers understand and represent your content, which in turn affects what the AI layer surfaces. Microsoft's Bing Copilot confirmed in March 2025 that schema helps their LLMs understand content structure. Google's Search team stated in April 2025 that structured data provides an indexing advantage.
Pages cited by AI systems are roughly three times more likely to have JSON-LD than uncited pages — but this is correlation, not causation. Technically sophisticated sites that maintain proper schema markup also tend to produce higher-quality content, earn more authoritative backlinks, and invest in all the other signals that drive AI citations. Schema is a marker of a well-maintained site, not the cause of the citations those sites earn.
B2B companies should implement schema in this order: first, Organization schema to establish entity clarity; second, FAQPage schema on any page with genuine Q&A content; third, Service schema on service or product pages; fourth, BlogPosting or Article schema on content pages. This sequence prioritizes the schema types with the clearest AI search benefit while building the foundational structured data layer that supports broader visibility.
We'll run 15 of your buyers' top queries across ChatGPT, Perplexity, and Google AI Overviews — and tell you exactly where you're appearing, where you're not, and what's driving the gap.
Book a Free AI Visibility CheckFree — No Commitment
We test 15 of your buyers' top queries across ChatGPT, Perplexity, and Google AI Overviews — and show you exactly where you appear, where competitors are being cited, and what's driving the gap.
Check My AI VisibilityTakes 2 minutes to request. We'll follow up within one business day.