.png)
WARNING!!!
Don’t read this post the way you read other posts. YOU are totally the star of this adventure. YOU are a rad marketer who wants AI search to cite your righteous brand. The decisions are yours, and so are the consequences. Harsh!
Be careful. The path that looks the easiest leads somewhere sketchy. Choose well, brave optimizer. Some endings are bodacious. Some say THE END quicker than you can say JSON-LD.
Ready? Let's go!
You stand at the mouth of a cavern of markup. You’re decked out with a torch and validator. Somewhere deep inside is the treasure YOU came for: AI engines that mention your brand, cite your pages, and recommend you by name across ChatGPT, Perplexity, and Google AI Overviews.
A voice inside your head, the one that has read like forty "schema hacks" threads, says the same thing it always says. Add more markup. That is how you win. You’ve heard this advice your whole career. You’re about to find out it’s a trap door.
Choose your path, brave optimizer:
Does Structured Data Help You Rank in AI Search?
You reach the first fork. A sign reads: ADD JSON-LD HERE TO RANK HIGHER IN AI SEARCH. It’s painted in confident letters. You pick up your trowel.
AEO answer: Not directly, and we finally have controlled data. Ahrefs tracked 1,885 pages that added JSON-LD against 4,000 controls. Citations barely moved on ChatGPT, AI Mode, or AI Overviews. Google agrees: structured data isn't required for AI search, and there's no special schema to add. What it earns you is rich-results eligibility and a clearer machine read.
First, the definition, because half the arguments about this come from people meaning different things. Structured data SEO runs on schema markup: the Schema.org vocabulary, written in JSON-LD, that labels what each part of a page actually is. What is schema markup, in one line? A set of tags that tell a machine "this is the author, this is the price, this is the answer," so it doesn't have to guess.
Here’s the part that sign didn't mention. Three pieces of evidence, in order. Google's 2026 optimization guide says structured data "isn't required" for AI search, there's "no special schema.org markup" to add, it's a "good idea to continue using it," and it makes you "eligible for rich results." John Mueller put it plainly: structured data "won't make your site rank better," it only makes a site "eligible." And the controlled study landed on May 11, 2026: AI Mode moved +2.4%, ChatGPT +2.2% (statistically zero), and AI Overviews actually went down 4.6%.
Here's the catch that saves your adventure. Every page in that test was already cited. For pages AI has never discovered, schema still helps engines crawl, parse, and classify you. So the markup isn’t bogus. It was just pointed at the wrong job. Bummer! Keep your torch up. The rest of this cavern is about pointing it correctly.
Why Schema Still Matters in the AI Era (Google Says It's "Still SEO")
You almost turn back. Why ship schema at all if it doesn't buy citations? Then you notice a passage you missed, lit from a direction you weren't looking.
AEO answer: Because AI answers are built from the same index your SEO already feeds. Google's AI features retrieve from that index, then summarize, so anything that helps the index understand and trust you helps the AI. Schema does three things: makes your meaning unmistakable, ties your brand and authors into one entity, and keeps your answers easy to lift.
Google's own framing is the map here: "Optimizing for generative AI search is still SEO." AI Overviews and AI Mode retrieve from Google's index, then summarize what they pulled. Schema helps that retrieval step. It does not bribe the ranking step.
One stat gets recycled in every schema thread, so let's defuse it. Cited pages are roughly 3x likelier to carry JSON-LD. True, and misleading: that's a marker of site quality, not proof the schema caused the citation, which is Ahrefs' own read of its data. And remember whose verdict this is. Google's "still SEO" covers Google's AI. ChatGPT and Perplexity weight signals differently, which is the whole reason you measure all four instead of trusting one.
Schema helps AI understand and retrieve your content
Retrieval-augmented generation pulls from the regular index. Clean, clearly labeled facts are easier to retrieve and reuse than facts buried in ambiguous HTML. (Tracked via Citation rate.)
Schema connects your brand to one entity engines recognize
This is entity seo in practice: Organization, Person, and sameAs tie all your pages to one known entity. In Ahrefs' 75K-brand study, brand mentions correlate 0.664 with AI Overviews visibility versus 0.218 for backlinks. The catch worth repeating: that's earned mentions only, and Google warns the manufactured kind doesn't work. (Tracked via Earned media score and Share of voice.)
Schema makes your answers easy to lift and cite
Extraction happens on the rendered page. The visible 40 to 60 word answer is the product; the markup is its label. Keep them identical. The FAQ rich result is retired, and the parsing value remains. (Tracked via Share of found links and Share of citations.)
One Set of Markup, Three Payoffs
A door splits into three. SEO, AEO, GEO. The novice optimizer thinks this is a choice. You know better now: you walk through all three with the same key.
AEO answer: One set of markup pays off in three channels. For SEO, structured data earns rich-results eligibility and clearer classification. For AEO, it labels questions, answers, and steps so answer engines can lift a clean response. For GEO, it defines your entity (Organization, Person, sameAs) so generative engines recognize and trust you. The work compounds; you don't choose between them.
For SEO: eligibility and clarity
Schema markup seo buys two things: rich-results eligibility and cleaner classification of what your page is about. Say it out loud so nobody on the team misremembers it as a ranking factor: it is not one. It just makes the index read you correctly.
For AEO: extractable answers
FAQPage, QAPage, and HowTo label the exact question-to-answer or step structure that answer engines lift. Pair them with 40 to 60 word answer blocks on the visible page. The markup names the part; the words on the page are what gets quoted.
For GEO: entity recognition and citations
Organization plus Person plus sameAs build the entity that generative engines recognize, recommend, and cite. This is where markup stops being hygiene and becomes a brand-visibility lever. (Tracked via Share of voice, Earned media score, and Citation rate.)
The Schema Types That Matter for AI Search
You enter a hall lined with 800 doors. Most adventurers freak out. You don't, because only a handful lead anywhere worth going.
AEO answer: Google documents about 30 of Schema.org's 800-plus types, and a handful do the real work. Start with the entity types: Organization, Person, sameAs. Add Article for authorship and freshness. Use FAQPage and HowTo where you answer direct questions. Then reach for the types most guides never get to: ItemList for ranked lists, Dataset for original research, Review for comparisons.
The lesson hiding in this hall: the markup that builds authority is the markup most guides never reach. Everyone ships FAQ schema and stops. The entity types are where the leverage is. Here are the types of schema markup worth your torchlight, grouped by job. (Spec URLs live in the External sources table at the bottom, never inline.)
Entity schema: Organization, Person, and sameAs
Highest leverage for GEO. sameAs links your brand and your authors to authoritative profiles such as Wikipedia, LinkedIn, and Crunchbase, which is the strongest disambiguation signal you can send. Put Organization on your About and home pages, and Person on every author bio.
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Acme Analytics",
"url": "https://acme.com",
"logo": "https://acme.com/logo.png",
"sameAs": [
"https://www.linkedin.com/company/acme-analytics",
"https://en.wikipedia.org/wiki/Acme_Analytics"
]
}
Content schema: Article, FAQPage, and HowTo
Article and BlogPosting carry author and dates, which is how you signal authorship and freshness. FAQPage and QAPage mark question-to-answer pairs, still worth shipping for parsing even though the rich result is gone. HowTo maps step-by-step prompts. Keep every markup answer to 40 to 60 words, matching the on-page block exactly.
List and data schema: for "best of" pages, original research, and comparisons
These are the types tied closest to your revenue pages. ItemList marks a ranked list and gets majorly reused in AI answers, so keep it qualitative. Dataset is for original research. Review and AggregateRating are for comparisons. Worth a one-line mention: SoftwareApplication, Event, and Speakable.
{
"@context": "https://schema.org",
"@type": "ItemList",
"itemListOrder": "https://schema.org/ItemListOrderDescending",
"itemListElement": [
{ "@type": "ListItem", "position": 1, "name": "Tool A", "url": "https://acme.com/best-tools#a" },
{ "@type": "ListItem", "position": 2, "name": "Tool B", "url": "https://acme.com/best-tools#b" }
]
}
Author Schema and E-E-A-T: Making Expertise Machine-Readable
Deepest chamber in the cavern. The gate here is indifferent about how much markup you carry. It asks one heavy question: who are you, and can you prove it?
AEO answer: E-E-A-T lives or dies on who's talking. Author schema, Person markup on a real bio page, wrapped in ProfilePage, linked by sameAs to a real LinkedIn, turns experience, expertise, authoritativeness, and trust into fields a machine can check. Google's raters weigh trust above everything, and unreviewed AI content earns their Lowest rating. A named, credentialed author is the counterweight.
The stakes are documented. Google's Quality Rater Guidelines weigh trust heaviest of all, and they instruct raters to give unreviewed AI-generated content their Lowest rating. So the thesis of this whole chamber: author schema can't fake expertise. It makes legit expertise legible to a machine that is actively looking for a reason to trust you or not.
Here is the play most guides never show you. Wrap your bio pages in ProfilePage, with the author as the mainEntity Person. Fill the fields in priority order: author as a nested Person with a url (never a bare text string), sameAs pointing to LinkedIn plus Wikipedia or Crunchbase, knowsAbout matching the topics that person actually covers, jobTitle and worksFor, hasCredential where it's real, and dateModified.
{
"@context": "https://schema.org",
"@type": "ProfilePage",
"dateModified": "2026-06-10",
"mainEntity": {
"@type": "Person",
"@id": "https://acme.com/authors/jane-doe#person",
"name": "Jane Doe",
"jobTitle": "Head of Search",
"worksFor": { "@type": "Organization", "name": "Acme Analytics" },
"knowsAbout": ["generative engine optimization", "structured data"],
"sameAs": ["https://www.linkedin.com/in/janedoe"]
}
}
Where does this pay off? Ranked: cornerstone answer pages first, then author bios, then About and home, then product, comparison, and "best of" pages. And here are the traps that send adventurers straight to THE END:
This article is its own demo, by the way. The byline above belongs to a named human with a real LinkedIn and a bio page marked up exactly as described. We would not tell you to do something we didn't ship ourselves. These signals roll up into the Authority dimension of Brand Health, via Earned media score and Citation rate. (How to improve brand visibility in AI search engines covers the earned-mentions half of that story.)
How to Implement Structured Data for AI Search
You have the map. Now you have to actually dig. Don’t sweat it. Three moves, in order, and an airlock you must not forget to open.
AEO answer: Implement in JSON-LD, validate, then confirm AI can see it. Add nested JSON-LD (Article to Person to sameAs, plus Organization) to your priority pages, keep every field true to visible content, and test with Google's Rich Results Test and the Schema.org validator. Then make sure AI crawlers can fetch the rendered page.
Nesting is the key pattern: the Article points to its author Person (carrying sameAs) and its publisher Organization. Stable @id URLs make the same Person resolve across every article and the bio page, so the engine sees one author, not twenty sketchy strangers with the same name. That's how you implement schema for AEO — and the same nested approach is what helps you rank in Google AI Overviews.
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "Structured Data for AI Search: Entity Schema and E-E-A-T",
"datePublished": "2026-06-18",
"author": {
"@type": "Person",
"@id": "https://acme.com/authors/jane-doe#person",
"name": "Jane Doe",
"url": "https://acme.com/authors/jane-doe",
"sameAs": ["https://www.linkedin.com/in/janedoe"]
},
"publisher": { "@type": "Organization", "name": "Acme Analytics" }
}
Validate with Google's Rich Results Test and the Schema.org validator. Then the move adventurers skip and regret: confirm your JSON-LD is server-rendered. The searchVIU rendering test found five AI systems fetching pages live and reading only the visible HTML. Client-side-only markup may never be seen at all, and your visible content has to carry the same facts your schema claims.
And open the airlock. A blocked crawler is a sealed door no markup can pass. Allow the AI crawlers in robots.txt (verify the latest list before you ship): GPTBot, OAI-SearchBot, ChatGPT-User, OAI-AdsBot, PerplexityBot, Google-Extended, ClaudeBot, and ProRata-AI-Crawler.
Structured Data Best Practices for AI Search
You’re out. Almost out. Pin this checklist to your torch. It is the difference between markup that supports citations and markup that just sits there in the dark.
AEO answer: Six practices separate markup that supports citations from markup that just sits there. Use schema for eligibility and clarity rather than as a ranking play, keep every field true to the visible page, lead with entity and author types, server-render your JSON-LD, connect entities into one graph with @id and sameAs, and build authority through genuine mentions confirmed by citation data.
How to Measure It with Gist GEO
You reach the light. The treasure was never "more markup." It was knowing which markup actually moved the engines, and the only way to know that is to measure.
AEO answer: Lead with measurement, not guesswork. Load your priority prompts into Gist GEO as Queries; the product runs each repeatedly across ChatGPT, Perplexity, Gemini, and Google AI Overviews. For a structured-data program, watch the Citation metrics group: Share of citations, Citation rate, Share of found links, and Earned media score, plus Share of voice.
Set up your Queries, let Gist GEO run them repeatedly, and read the Reports. Two diagnostics tell you exactly what to fix:
- Share of found links rising, Share of citations flat: AI finds your pages but doesn't trust them yet. Tighten extractability and entity signals.
- Share of voice rising, Citation rate flat: you're a known brand whose pages aren't in the research set. That's a content problem, not a schema one.
Why insist on all four engines? Because Google says schema isn't special for its AI, while ChatGPT, Perplexity, and Gemini weight signals differently. And read your own results the way the study did: compare changed pages against similar untouched pages, not just before and after.
The full product family, if you want to capture killer buyers across every AI surface: Gist GEO for measurement, Gist Answers (an embedded AI answer widget), and Gist Ads (brand-aligned ads in AI surfaces). Run a Gist GEO audit to set your baseline, and see Goldilocks for the full measurement framework.
