Short answer
Structured data for AI search is schema.org markup (usually JSON-LD) that labels your entities so machines read them without guessing. The types worth implementing first are Organization, Article, LocalBusiness, and FAQPage. Schema helps Google understand entities and can trigger rich results, but there is no public evidence it directly boosts citations in ChatGPT, Perplexity, or AI Overviews. Treat schema markup for ChatGPT and Google as a clarity layer, not a ranking trick. Implement Organization first, validate with Google's Rich Results Test and the Schema Markup Validator, then keep it accurate. Expect cleaner entity understanding and eligibility for rich results, not guaranteed placement or a citation spike.
What structured data is and why it matters for AI search
Structured data for AI search is schema.org markup, usually written as JSON-LD, that labels the entities on a page so machines read them directly instead of inferring them from prose. It names your organization, an article, a product, or a physical location in a vocabulary that search systems already parse (see the vocabulary at schema.org). Google recommends JSON-LD and states that valid markup makes a page eligible for enhanced appearances, per its structured data basics documentation.
It matters for AI search because AI engines lean heavily on Google's index and on entity understanding. Clean markup is one signal that helps a system confirm who you are and what a page is about. The honest caveat, which we repeat throughout: there is no public evidence that schema markup directly increases citations in ChatGPT, Perplexity, or Google's AI Overviews. The mechanism is indirect. Structured data helps Google understand entities and can trigger rich results; from there, better entity clarity may help downstream systems, but no vendor documents a direct schema-to-citation boost.
Which schema.org types actually surface in AI answers
The types worth your time are Organization, Article, LocalBusiness, and FAQPage, because they map to entities AI systems already reason about. Google's full list of supported types lives in its structured data feature gallery, and the vocabulary itself is defined at schema.org.
- Organization — establishes your brand as an entity: name, logo, URL, sameAs profiles. This is the single most useful type for entity recognition.
- Article — labels headline, author, and publish date, which supports authorship and freshness signals.
- LocalBusiness — adds address, hours, and geo data for businesses with a physical presence.
- FAQPage — structures question-and-answer content, though its visible payoff has narrowed sharply (see the myths section below).
Two reality checks. FAQ rich results are now limited to well-known authoritative government and health sites, per Google's August 2023 update. HowTo rich results were removed entirely, per Google's September 2023 announcement. Markup for those types can still describe your content, but do not expect the visible feature.
How to implement schema markup with JSON-LD
Implement schema in JSON-LD, in a defined order, starting with the entity that anchors everything else: your organization. Google recommends JSON-LD as the format in its structured data basics guide, because it sits in a script block and does not tangle with your visible HTML.
- Step 1 — Organization, site-wide. Add it to your homepage or a global template so your brand is defined once, cleanly.
- Step 2 — Article, on every editorial page. Template it so headline, author, and dates populate automatically.
- Step 3 — LocalBusiness, if you have a physical location. Keep name, address, and phone identical to your other listings.
- Step 4 — FAQPage, only where genuine Q&A exists. Do not manufacture questions to earn markup.
Match every property to visible on-page content; Google's documentation is explicit that structured data must represent what users actually see. Generate it from your data layer rather than hand-writing it per page, so it stays accurate as content changes.
Schema markup testing tools and validation workflow
Validate every schema change with two tools before and after deployment: Google's Rich Results Test and the Schema Markup Validator. Use the Rich Results Test to confirm which Google features your markup is eligible for, and the Schema Markup Validator to check that your JSON-LD is valid against the schema.org vocabulary regardless of Google features.
A workflow that holds up in production: draft the JSON-LD, run it through both validators, fix errors and warnings, deploy, then re-test the live URL to confirm the rendered markup matches what you wrote. In our experience running these engagements, the most common failure is markup that validates in a test tool but never renders live because it is injected after page load in a way crawlers do not reliably capture. Always test the deployed URL, not just the code snippet.
Structured data myths (schema is not magic)
Three myths cause most wasted effort: schema guarantees rich results, more schema means more rankings, and FAQPage markup still wins featured Q&A slots.
- Myth: valid markup guarantees a rich result. Google's own structured data documentation states that markup makes a page eligible for enhanced appearances but does not guarantee them, and that Google decides whether to show a feature — eligibility, not entitlement.
- Myth: adding more types lifts rankings. No Google documentation presents structured data as a ranking factor; the same structured data basics guide frames it as an aid to understanding and eligibility, nothing more.
- Myth: FAQPage still earns visible Q&A features broadly. Those results are now restricted to authoritative government and health sites, per the August 2023 FAQ update, and HowTo results were removed entirely in the September 2023 announcement.
The biggest myth of all is that schema is a shortcut to AI citations. It is not. Structured data clarifies entities; it does not write the substance an AI engine chooses to quote. For where citations really come from, see Get cited by ChatGPT.
What does structured data cost
Structured data costs mostly time, not tooling, because the core validators are free. The paid inputs are engineering hours to template it and ongoing maintenance to keep it accurate. The ranges below are labeled as experience-based, not benchmarks.
- Tooling: near zero. The Rich Results Test and Schema Markup Validator are free; most CMS platforms output basic schema without a plugin purchase.
- In-house time. In our experience running these engagements, a first, correct pass across Organization, Article, and LocalBusiness typically takes a developer a few days on a standard site, longer on a large or templated one.
- Agency vs in-house. An agency buys you speed and fewer validation mistakes; in-house saves cash if a developer already understands your data layer. Neither changes the ceiling on what schema can do.
- Maintenance. Budget recurring time whenever templates, authors, or business details change, so markup never contradicts the visible page.
LYVIA is a Paris-based AI agency serving international clients, so we scope schema work to fit lean US and UK teams rather than selling a giant retainer for a job that is mostly one-time.
Structured data vs content: where AI visibility really comes from
AI visibility comes from substantive, quotable content far more than from markup — schema clarifies your entities, but it cannot invent the answer an engine decides to cite. Structured data and content are complementary layers: markup tells a machine what an entity is, while content gives the machine something worth repeating.
Think of schema as the label and content as the product. A perfectly labeled page with thin content still has nothing to quote. This is why we sequence engagements so schema is a quick foundation, then invest the bulk of effort in the writing and structure that AI engines actually surface. For the content side, see SEO and GEO content strategy and, for the Google feature specifically, Appear in Google AI Overviews. If you want the broader framework, What is generative engine optimization sets the context this technical piece sits inside.
Frequently asked questions
Does schema markup help you get cited by ChatGPT or Perplexity?
There is no public evidence that schema markup directly increases citations in ChatGPT, Perplexity, or Google's AI Overviews. The mechanism is indirect: structured data helps Google understand your entities and can make pages eligible for rich results, and clearer entity understanding may help downstream systems. But no vendor documents a schema-to-citation boost, so treat any claim of a direct lift skeptically. Implement Organization and Article schema for clean entity signals, then invest most of your effort in substantive, quotable content, which is what AI engines actually surface and repeat.
Which schema types should a small business implement first?
Implement Organization first, then Article, then LocalBusiness if you have a physical location, and FAQPage only where genuine question-and-answer content exists. Organization anchors your brand as an entity with name, logo, URL, and sameAs profiles, which is the highest-value signal for entity recognition. Article supports authorship and freshness on editorial pages. LocalBusiness adds address and hours for physical locations. FAQPage is last because its visible rich result is now limited to authoritative government and health sites, so add it for structure, not for an expected feature. Validate each with Google's Rich Results Test.
Is JSON-LD better than microdata for structured data?
JSON-LD is the recommended format in Google's structured data documentation, and it is the practical choice for most teams. It lives in a script block separate from your visible HTML, so it is easier to generate from a data layer, easier to template, and less likely to break when you redesign a page. Microdata and RDFa still work, but they interleave with your markup and are harder to maintain at scale. Write JSON-LD, generate it from your content source rather than by hand, and make sure every property matches what users actually see on the page.
Will structured data guarantee a rich result in Google?
No. Google's structured data documentation states that valid markup makes a page eligible for enhanced appearances but does not guarantee them; Google decides when to display a feature. That is eligibility, not entitlement. Some features have also narrowed: FAQ rich results are now restricted to well-known authoritative government and health sites as of Google's August 2023 update, and HowTo rich results were removed entirely in September 2023. So implement schema to describe your content accurately and to become eligible, but plan your strategy around clearer entity understanding rather than any promised visual feature.
How do you test and validate schema markup?
Use two free tools together: Google's Rich Results Test to confirm which Google features your markup is eligible for, and the Schema Markup Validator to check that your JSON-LD is valid against the schema.org vocabulary. The reliable workflow is to draft the markup, validate the snippet in both tools, fix errors and warnings, deploy, then re-test the live URL to confirm the rendered markup matches your code. Testing the deployed page matters because markup that is injected after load can validate as a snippet yet fail to render for crawlers. Re-validate after any template or content change.
LYVIA is a Paris-based AI agency that helps US and UK small businesses get the technical foundations right without overselling what schema can do. If you want structured data implemented cleanly, validated, and paired with content that AI engines actually cite, Book a call
