Humans read prose; machines prefer structured facts. Structured data — primarily JSON-LD using the schema.org vocabulary — lets you state explicitly what a page is (an Article, a Product, an FAQ), who stands behind it (an Organization or Person, with verified profiles), and what it claims. That removes guesswork for search engines and, increasingly, is a primary input for AI answer engines deciding what to cite.
Each check has a stable ID (D1–D9). The first five are core; the rest are extended, type-specific markup.
Structured data is markup that describes your content to machines in a standard vocabulary (schema.org). The recommended format is JSON-LD — a <script type="application/ld+json"> block in the page — which Google prefers over the older inline Microdata and RDFa because it's clean and decoupled from the visible HTML.
Its original payoff was rich results: star ratings, FAQ accordions, product prices and breadcrumbs in the SERP, which lift click-through. But the bigger shift is GEO: AI Overviews, ChatGPT Search and Perplexity use structured data as a trusted, unambiguous description of what a page is and claims, which feeds directly into what they choose to quote.
A page with no structured data forces engines to infer everything from prose. Adding even a basic, accurate JSON-LD block for the page's type is one of the highest-leverage GEO investments available.
Structured data only helps if it's correct. Each block must be syntactically valid JSON — a single misplaced comma and the whole block is silently ignored. And it must use an appropriate `@type`: an Article for an article, Product for a product, FAQPage for an FAQ — with the recommended properties for that type filled in.
Validate against Google's Rich Results Test and the Schema.org validator. Two failure modes are common: a syntax error that kills an otherwise-good block, and a technically-valid but wrong or skeletal type that adds no real information.
Critically, structured data must reflect the [visible content](category-content.html). Marking up an FAQ that doesn't appear on the page, or a rating that isn't real, is a guidelines violation that can earn a manual penalty.
Entities are how the modern web reasons about who. An Organization or Person schema block, with a name, logo and canonical URL, declares the publisher. The key property is `sameAs`: an array of links to your verified profiles elsewhere — Wikidata, Wikipedia, LinkedIn, Crunchbase, official social accounts.
Those sameAs links let an engine disambiguate and trust your entity by connecting it to known references. This is foundational for GEO: AI engines build their answers around entities they can resolve, and they're far more likely to cite a clearly-identified, well-connected publisher than an anonymous page. It works hand-in-hand with the authorship and entity signals in GEO / LLM readiness.
An entity block without sameAs is a start; an entity block with verified profile links is what builds the trust graph that AI engines reward.
Three schema types form the backbone that most sites should always include. Organization establishes the publisher (and pairs with sameAs). WebSite describes the site itself (and can enable a sitelinks search box). BreadcrumbList exposes the page's position in the site hierarchy, which shows as a breadcrumb trail in results and helps engines understand structure.
Together these give engines a consistent, machine-readable picture of who you are and how the site is organized — context that benefits every page, not just the one being viewed.
This check looks for the presence of these foundational types; missing all three means engines are working with less context than competitors who include them.
Open Graph is the metadata that builds the preview card when your URL is shared — on social platforms, in chat apps, and in various link-unfurling contexts. The essentials are og:title, og:description, og:image, og:url and og:type.
Without them, shared links render as a bare URL or an auto-picked, often-wrong image — which kills the click-through on social. With them, you control the headline, summary and image of every share.
It's a small, one-time addition per template that pays off everywhere your links travel, and it reinforces the title/description/image signals engines already read.
Extended checks
Diagnostic checks flagged for review — useful polish, but they don't carry the same score weight as the core checks above.
A BreadcrumbList describes the trail from the homepage to the current page (Home › Category › Page). Marked up in JSON-LD, it can replace the raw URL in search results with a readable breadcrumb, and it reinforces how the page fits into the site's structure.
If you already show visual breadcrumbs, adding the schema is a quick win; if you don't, breadcrumbs improve both UX and structural clarity.
These types describe answer-shaped content. FAQPage marks up question-and-answer pairs, QAPage a single Q&A, and Article a news or blog article (with headline, author and dates). They make the question-answer structure explicit, which both enables rich results and gives AI engines clean, pre-segmented answers to lift.
Use the type that genuinely matches your content, and ensure the marked-up Q&A actually appears on the page — invented FAQ markup is a guidelines violation.
On product and service pages, Product schema (name, description, price, availability) and Review/AggregateRating schema unlock some of the most valuable rich results — prices and star ratings right in the SERP — and feed comparison and recommendation answers in AI engines.
Mark up real, accurate data only. Fake or mismatched review markup is a common cause of manual penalties.
Twitter/X reads its own twitter:card meta tags (falling back to Open Graph) to render share previews. Declaring twitter:card (ideally summary_large_image) plus title/image gives you control of how links look on X.
It's optional and overlaps with Open Graph, but it's a quick addition that improves one more surface where your links appear.
Why structured data matters for SEO and GEO
Structured data is the bridge between human-readable content and machine understanding. For classic SEO it earns rich results — the prices, ratings, FAQs and breadcrumbs that make your listing stand out and lift click-through. But its role has been quietly upgraded: it is now core infrastructure for AI answer engines, which use it as a trusted, unambiguous description of what your page is, who published it, and what it claims.
The highest-value pieces are the entity layer — a clear Organization/Person with sameAs links to verified profiles — and accurate, type-appropriate page markup. Together they tell an AI engine exactly who you are and what you're saying, which is precisely what it needs to cite you with confidence.
AI Overviews, ChatGPT Search and Perplexity treat structured data as a trusted, pre-digested description of your page — so it's no longer just "rich-result candy," it's how you become quotable. The entity layer matters most: a clear Organization or Person with sameAs links to Wikidata, LinkedIn and your official profiles is how an answer engine resolves who you are and decides you're a credible source worth citing.
Audit your page across all 9 checks in Structured Data
See exactly which checks pass, warn or fail — in seconds.
Run a free audit →