Schema markup for AI search, explained
A priority-ordered breakdown of the schema types that influence AI search visibility, how engines use them, and how to validate your markup.
By ShowUp AI · Published
Schema markup is structured data you add to your pages, usually as JSON-LD, that states facts explicitly instead of leaving AI systems to infer them from prose — who you are, what you sell, what a page answers, and how your site is organised. For AI visibility, Organization, Product, FAQPage, Article, BreadcrumbList and LocalBusiness are the types that matter most, and they're worth doing in that rough order.
Here's what each type does, how AI engines actually use it, how to check your markup is valid, and where to start if you can only do a few pages this month.
Why does schema markup matter for AI search specifically, not just SEO?
Answer first: AI systems generating an answer need to extract facts quickly and confidently, and structured data removes the guesswork — a model reading clean JSON-LD can pull your price, your FAQ answer or your company details directly rather than trying to parse them out of paragraphs, headers and page layout. That confidence is part of why a source gets cited or a brand gets named.
A few mechanics worth understanding:
- It reduces ambiguity. A price stated in a sentence ("plans start at a competitive rate") is useless to extract; a Product schema with a numeric price and currency is not.
- It supports entity resolution. Organization schema with a consistent name, URL, logo and sameAs links to your social/Wikidata/Crunchbase profiles helps a model tie mentions of you across sources back to one confident entity.
- Google's AI Overviews are documented to draw on structured data, particularly FAQPage, HowTo (where still supported) and Product markup, alongside normal page content.
- Other engines are less transparent but directionally similar. ChatGPT search and Perplexity don't publish exactly how they weight schema, but well-structured, factual pages consistently perform better in visibility tracking than pages saying the same thing in loose prose — schema is a big part of what makes a page "well-structured."
Schema doesn't guarantee a mention. It raises the odds that if your page is retrieved, the facts on it get extracted correctly and attributed to you rather than left out or garbled.
Which schema types actually matter, and what does each one do?
Answer first: for most brands, six types cover nearly everything worth doing — Organization for who you are, Product for what you sell, FAQPage for direct question-answer content, Article for blog/editorial content, BreadcrumbList for site structure, and LocalBusiness if you have a physical or service-area presence. Everything else is a smaller add-on for specific situations.
| Schema type | What it establishes | Where to use it |
|---|---|---|
| Organization | Brand identity: name, logo, URL, social profiles, founding info | Homepage, about page |
| Product | Name, price, availability, reviews for a specific offering | Product/pricing pages |
| FAQPage | Explicit question-answer pairs | Any page with genuine FAQs |
| Article | Author, publish date, headline for editorial content | Blog posts, guides |
| BreadcrumbList | Site hierarchy and page relationships | Any page with a breadcrumb trail |
| LocalBusiness | Address, hours, service area, phone | Location or contact pages |
Organization
This is your entity's core identity record. Include legal/brand name, url, logo, a short description, sameAs (links to LinkedIn, Crunchbase, Wikidata, X, YouTube — whichever apply), and contactPoint if relevant. Put it once, on your homepage, and keep every field matching what's on your other listings — mismatches undercut the entity-resolution benefit rather than just being a cosmetic issue.
Product
If you sell a defined product or plan, mark up its name, description, price (or price range), currency, availability and, if you have them, aggregate ratings. This is the type most directly tied to "how much does X cost" and "is X worth it" questions — exactly the kind of question people ask AI assistants before they'd ever visit your site.
FAQPage
Mark up genuine FAQs — real questions your buyers ask, with complete, self-contained answers, not marketing copy dressed as a question. This is the type with the clearest documented link to AI Overviews specifically, and it maps directly onto the kind of question-and-answer pairs that assistants are built to reproduce. Don't mark up decorative accordions with vague non-answers; a shallow FAQPage schema on unhelpful content doesn't help and can look manipulative to reviewers if Google spot-checks it.
Article
For blog posts and guides, headline, author (a real named person, not "Admin"), datePublished, dateModified and publisher. This supports freshness signals and gives assistants a byline to attribute if they quote you — increasingly relevant as engines get more careful about sourcing claims to identifiable authors rather than anonymous pages.
BreadcrumbList
Low effort, often auto-generated by CMS platforms or plugins. It clarifies where a page sits in your site's structure, which helps a crawler understand context (this is a pricing page under "Product," not a random landing page). Minor on its own, but it's close to free if your CMS supports it natively.
LocalBusiness
If you have a physical location or a defined service area, this is what feeds Google's local and Maps-adjacent AI answers: address, phone, hours, geo-coordinates, service area, and price range. For multi-location brands, each location needs its own instance rather than one shared block.
Worth knowing about but lower priority
- Review/AggregateRating — valuable but only mark it up if reviews are genuine and visible on the page; Google has taken action against fabricated or hidden ratings markup.
- HowTo — support has been scaled back in some Google surfaces; still fine to add but don't expect much from it currently.
- VideoObject — matters more if video is a core content format for you, particularly for Gemini/YouTube-influenced answers.
How do AI engines and Google actually use this data in practice?
Answer first: Google's AI Overviews and standard search both read structured data as part of how they understand a page well enough to feature it, and the same clean structured facts that help Google are generally the same facts that make a page easier for any assistant's retrieval and summarisation step to use correctly — even engines that don't officially document schema support benefit indirectly because well-marked-up pages tend to be well-organised pages generally.
A few realistic caveats to keep expectations honest:
- Schema is a signal, not a ranking guarantee. It doesn't override thin content, poor authority, or the absence of any real reason to cite you.
- Not every engine documents its use of schema. OpenAI and Perplexity haven't published detailed statements on how heavily they weight structured data versus prose content; the practical evidence is indirect — well-structured pages tend to be cited more, and schema is one input into "well-structured."
- Rich results and AI citations are different things. Getting a rich snippet in classic Google search (stars, FAQ dropdown) doesn't automatically mean an AI Overview will cite you, though the underlying markup helps both.
- Freshness fields matter. dateModified being genuinely current, not just bumped by a script, correlates with being treated as an authoritative, maintained source.
How do you check your schema is actually valid?
Answer first: use Google's Rich Results Test and the Schema.org validator to check syntax and required fields before you assume markup is working, then spot-check with your browser's "view page source" or a rendering tool to confirm the JSON-LD is actually present in what gets served, not just in your CMS editor.
A simple validation workflow:
- 1Rich Results Test (Google). Paste the live URL. It flags missing required fields and tells you which rich result types Google recognises the page for.
- 2Schema.org / Schema Markup Validator. More permissive and type-complete than Google's tool; useful for catching structural errors Google's test doesn't flag because it doesn't care about that type.
- 3View source, not just the CMS preview. Some CMS and page-builder plugins render schema client-side via JavaScript, which some crawlers may not execute reliably — check the raw HTML response, not just what renders in a browser.
- 4Re-check after every redesign or CMS migration. Schema is exactly the kind of thing that silently breaks when a theme or plugin changes, and nobody notices because the page still looks fine to a human visitor.
- 5Keep one reference JSON-LD block per type so new pages start from a validated template instead of someone free-handing markup from memory.
A minimal, valid Organization block looks like this in plain text (add real values and wrap it in a script tag of type application/ld+json on your page):
type: Organization name: your company name url: your homepage URL logo: URL to your logo sameAs: [ your LinkedIn URL, your Crunchbase URL, your X URL ]
If you can only do a few pages, what's the priority order?
Answer first: start with Organization on your homepage, then FAQPage on your highest-traffic FAQ or pricing content, then Product on your core offering pages — those three cover identity, direct Q&A extraction and purchase-decision facts, which are the three things AI assistants get asked about most.
A realistic rollout sequence:
- 1Organization schema, homepage. One-time setup, highest leverage for entity resolution.
- 2FAQPage on your best FAQ content. Pick the page that already answers real buyer questions, not the thinnest one.
- 3Product/pricing schema on your core offering. Directly answers "how much does X cost" queries.
- 4Article schema across blog/guides, ideally templated so every new post gets it automatically.
- 5BreadcrumbList site-wide, usually a CMS/plugin toggle rather than manual work.
- 6LocalBusiness, if applicable, per location.
- 7Review/AggregateRating, only once you have genuine, visible reviews to back it.
Schema markup is one part of a broader AI visibility setup alongside listings, content structure and monitoring — for the fuller picture of what to prioritise across all of it, see the AI visibility checklist. If you want to see how your current pages and structured data stack up against competitors right now, the free AI visibility check is a fast way to get a baseline before you start marking up pages.
Get named in AI answers — starting today
The $299 ShowUp AI Setup checks your live site and builds your report, technical kit (llms.txt, schema, meta, FAQs) and ready-to-publish content — with a 30-day re-check included. Monthly Watch plans are optional afterwards.
Related articles
How to Show Up in ChatGPT, Perplexity, Gemini & Google AI Overviews in 2026
Crawler access, answer-first content and consistent entity data are what decide whether an AI assistant names your brand. Here is the whole system, plus a 30-day plan.
Read itAEO vs SEO: What's the Difference and Why You Need Both in 2026
SEO gets you into the set of sources an AI engine considers. AEO decides whether it quotes you. A side-by-side comparison and a priority order.
Read itWhy Your Brand Is Invisible to AI Assistants (and How to Fix It)
AI assistants do not dislike your brand — they cannot find a sentence about you worth repeating. The five causes, how to confirm each, and the fixes by impact.
Read it