Ten Schema Tags That Get You Cited by AI
Schema markup alone does nothing — Ahrefs tested 1,885 pages and adding it with no content change produced no citation lift. But Organization and FAQPage JSON-LD on top of real content earn 2.7x more AI citations. Here's the code and the deployment checklist.
There's a tempting shortcut in every GEO discussion: add structured data, get cited more. It's tempting because it's technical, controllable, and finishable — the opposite of the messy earned-media work the last two pieces argued actually moves the needle. Write some JSON-LD, deploy it, check the box. And it does help. But it helps in a very specific way that's easy to get backwards, so start with the caveat that keeps you honest.
Schema is a multiplier, not a source
Ahrefs ran a controlled experiment across 1,885 pages: they added schema markup while holding content quality constant, and measured the effect on AI citations. The lift was not statistically significant. Adding structured data to a page whose content didn't change did essentially nothing for how often AI engines cited it.
Read that carefully, because it's the single most important sentence in this piece. Schema markup with no real content behind it produces no citation lift. It is not magic. It doesn't make a thin page citable. What it does — when there's substance underneath — is make that substance legible to a machine that reads the web as a graph of entities and relationships rather than a pile of prose. Schema is the label on the box. It helps the AI understand what's inside. It cannot put anything in the box.
Schema is the label on the box, not what's in it. Label an empty box and you've labeled an empty box.
With that fixed, two tags carry most of the weight for a solo founder. Get these two right, on top of genuinely useful content, and you've captured the majority of the available lift.
Tag one: Organization — how the AI learns who you are
An AI engine doesn't read your site as a list of keyword-matched pages. It builds an internal model of entities — a node for "your company," connected to what you do, who you serve, what you know about. Organization schema, placed in your homepage <head>, is how you hand the machine that node directly. Without it, your other pages are orphaned documents to a graph-based reader — the AI can see them, but it has no anchor connecting them to a coherent "you."
This is the foundation of the whole GEO setup, which is why it goes on the homepage and why the description and knowsAbout fields matter most. Write them in plain language — what you actually do, and the topics you're genuinely knowledgeable about — not a keyword-stuffed word salad. The machine is trying to understand you, not count your keywords.
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Acme Invoicing",
"url": "https://acme-invoicing.com",
"description": "Recurring invoicing and VAT handling built for solo founders and freelancers.",
"knowsAbout": [
"recurring invoicing",
"VAT compliance for freelancers",
"payment reconciliation"
],
"sameAs": [
"https://www.linkedin.com/company/acme-invoicing",
"https://github.com/acme-invoicing"
]
}
Drop that in a <script type="application/ld+json"> block in the homepage <head>. The sameAs array links your entity to your other verified profiles, corroborating that you're a real, consistent presence across the web — the same corroboration logic that governs AI citation generally.
Tag two: FAQPage — the highest ROI single step
FAQ content is the format AI engines cite most, because its structure — a real question paired with a direct answer — is exactly the shape of what a user asks and what the AI wants to return. The numbers are strong. Relixir's 2025 data found pages with FAQPage markup get cited 41% of the time versus 15% without — roughly 2.7x. PresenceAI measured an even larger effect, up to an 89% lift in citations.
But — and this is the caveat again, in miniature — the schema only works if there's a real FAQ on the page. Tagging an empty page does nothing; the answers have to actually be there. And they have to be answers to questions people really ask, not questions you wish they'd ask. Pull them from:
- Your actual support tickets and emails — the questions real users send you.
- Reddit and forum threads about your category — the questions people ask in public.
- Google autocomplete — start typing your topic and watch what completes.
Write 4–6 questions per core page, each with a 2–4 sentence answer in natural language. Not marketing copy — plain, direct answers a person would actually find useful.
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "Does Acme Invoicing handle VAT for EU freelancers?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes. Acme applies the correct VAT rate per client country, flags reverse-charge cases, and produces VAT-ready invoices you can export for filing."
}
},
{
"@type": "Question",
"name": "Can I set up recurring invoices without a monthly plan?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes. Recurring schedules are available on the free tier for up to three active clients, so you can automate billing before committing to a paid plan."
}
}
]
}
A bonus: FAQPage schema also helps you win Google Featured Snippets on the classic search path. One piece of work, both paths from the forked search landscape — the traditional list and the AI answer both reward it.
The full deployment checklist
Two tags carry most of the weight, but stacking multiple schema types covers more of the different ways AI engines parse pages. Digital Applied's analysis of 6.8 million AI citations found that pages stacking multiple schema types get cited 3.1x more than single-schema pages. Here's the minimal viable stack, in priority order:
| Schema type | Where it goes | Priority |
|---|---|---|
| Organization | Homepage <head> | Must |
| FAQPage | Every page with real Q&A content | Must |
| Article / TechArticle | Every blog post or guide | Must |
| BreadcrumbList | All inner pages | High |
| BlogPosting | Blog index and posts | Medium |
Don't stack schema for its own sake — each type should describe content that genuinely exists on the page. A BreadcrumbList needs a real navigation hierarchy; an Article needs a real article. The 3.1x figure comes from pages where the stacked schema reflects genuinely rich content, not pages that bolted on empty tags to game the count.
Validate before you trust it
Malformed JSON-LD fails silently — the AI just ignores it, and you never know. Run every page you mark up through a validator, and run two, because they catch different issues:
- Google's Rich Results Test — paste the URL or the code snippet; green check passes, yellow is a warning, red is an error. This checks Google-specific requirements.
- schema.org's own validator — checks conformance to the schema.org standard itself, which has slightly different rules. Worth running both; each surfaces problems the other misses.
The one thing to remember
Schema is worth doing. It's a one-time technical task that makes real content legible to machines, and the two tags above plus the stack will genuinely lift how often you're cited — on top of substance that earns citation on its own merits. But keep the order straight, because getting it backwards wastes the effort entirely:
- Write the real content first. A page has to be genuinely useful before any markup helps it. The Ahrefs experiment is definitive: schema without content quality lifts citations by zero.
- Organization on the homepage is the foundation. It's how the AI builds an entity node for you at all. Everything else is orphaned without it.
- FAQPage is the highest-ROI single addition — 2.7x citation rate — but only with real answers to real questions behind it.
- Stack multiple schema types for the 3.1x compounding effect, but only where each type describes content that actually exists.
- Validate on two tools before trusting any of it; silent failures are the default.
Schema is the label on the box. It's a good label, and a well-labeled box gets found faster. But the next piece is about what actually goes in the box at scale — and why a structured cluster of real content beats a single well-tagged page every time.
Part of a 5-piece look at GEO — getting cited by AI search engines. Previously: Search Traffic Forked: SEO and GEO Are Different Games, Why AI Engines Cite Reddit Over Your Homepage, and The Reddit Playbook for Solo Founders Who Hate Marketing. Next, the last piece: Pillar Pages and Content Clusters for AI Citation. More builder insights.