Schema markup for AI visibility: the complete guide

Schema markup: a language for machines

Schema markup (structured data) is JSON-LD code added to your pages' HTML to explicitly tell machines what you are : a business, a product, an article, a question-and-answer pair. It isn't content visitors see — it's a shared vocabulary that Google, Bing, and AI engines (ChatGPT, Perplexity, Gemini, Copilot) read to understand a page without having to guess.

Two different audiences get a different benefit from it : Google can display rich results in search (stars, price, FAQ — with important caveats, see below) ; an AI agent can extract a precise answer (price, average rating, address) without interpreting your page layout.

Six schema types, six different jobs

The question is never "which schema is best" — it's "which schema matches what this page actually describes." Here are the six types that cover nearly every case for a business, across any industry.

Organization

The schema that describes your business itself : legal name, logo, links to your official profiles (sameAs — social networks, Wikipedia, professional directories). Usually placed once, on the homepage or globally, and acts as an anchor : it's what Google and AI systems tie back to your brand everywhere else on the web.

LocalBusiness

For any business with a physical address and hours : full address, geolocation, openingHoursSpecification, priceRange, phone number. It's the parent type that dozens of industry-specific subtypes inherit from (restaurant, medical practice, hair salon, agency…).

Running a hotel or campsite ? The Hotel, LodgingBusiness, Campground and Resort subtypes have their own specific fields (starRating, amenityFeature, pets allowed…) — that's the subject of our dedicated guide : Schema markup for hotels and campgrounds.

Product

For anything you sell : price, availability, identifiers (GTIN, MPN), reviews. It's the schema Google Shopping has read for years — and the one the new AI shopping agents (ChatGPT Shopping, Perplexity Shopping) use to understand a catalog.

Selling on WooCommerce ? The stakes there are specific (product variations, incomplete native GTIN, syncing with merchant feeds) — see our dedicated guide : Product schema markup for AI shopping agents.

Article

For your blog and news pages : headline, author, datePublished, dateModified, image. Tells Google and AI systems this is dated editorial content, not a static commercial page — useful for anything tied to content freshness.

Review and AggregateRating

For displaying an average rating and review count. Google's rule is strict : only genuine third-party reviews count — you can't mark up your own marketing copy as a "review," and self-serving reviews of your own product have been explicitly excluded from Google's quality guidelines for years. An aggregateRating without verifiable reviews behind it is a liability, not an asset.

FAQPage

Here's an important update : since May 2026, Google no longer shows any FAQ rich result in search — eligibility, already narrowed to government and health sites back in August 2023, has now been fully withdrawn. If your goal is a Google rich snippet, FAQPage doesn't produce one anymore, for anyone.

What's still true : FAQPage remains a valid schema.org type, and a well-structured FAQ page (clear question/answer pairs, no guesswork) is still easier for an AI agent to extract a precise answer from — independent of any rich result. Keep FAQPage if you have genuine frequently asked questions; stop adding it hoping for a Google snippet that no longer exists.

Required vs. recommended fields

Common to every type : @type, @context, name. Without these, there's no valid schema.

Specific to each type : a LocalBusiness needs a complete address and a telephone ; a Product needs an offers with price and availability ; an Article needs a datePublished. The field most often missing, across every type : image — plenty of themes omit it by default.

Validate your schema in 2 minutes

Google provides a free testing tool : search.google.com/test/rich-results. Enter a page's URL and the tool flags what's valid, what's missing, and what's eligible for a rich result (accounting for retirements like FAQPage's). The generic validator.schema.org validator covers the types Google doesn't test.

Livada SEO automatically generates and validates the schema that fits your type of business — Organization and LocalBusiness by default, plus 18 specialized schemas depending on your pages (Product, Article, FAQPage, Review, and the hospitality-specific types).