Digi-Pilot
by Digitalix.be
← All articles

11 September 2026

JSON-LD Structured Data: The Markup That Makes You Quotable by AI

JSON-LD Structured Data: The Markup That Makes You Quotable by AI

A web page says a lot to a human reader and very little to a machine. A price, an address, an author, a last-updated date: all obvious at a glance, but a search engine has to guess them from raw text. Structured data exists for exactly this — translating your page’s content into explicit information, in a format Google, Bing and AI answer engines read without ambiguity.

What structured data actually is

Structured data is a block of code, invisible to visitors, added to the page’s HTML, describing its content through a shared vocabulary: Schema.org. The format Google recommends is JSON-LD: a script placed in the <head> or <body>, kept separate from the visible content.

In practice, instead of letting an engine infer that a page is about a company based in Brussels, open 9am to 5pm, you declare it:

  • @type states the nature of the thing described (Organization, Article, Product, LocalBusiness, FAQPage…)
  • properties (name, address, openingHours, author, datePublished…) carry the values
  • a single block can hold several linked entities, usually grouped in a @graph

One rule above all: the markup must describe what the page actually shows. Declaring a 5-star rating or a price that appears nowhere isn’t optimisation, it’s grounds for a manual penalty.

What it changes for classic SEO

Structured data is not a direct ranking factor — Google has said so repeatedly. Its effect is indirect, but very real:

  • Rich results. Review stars, breadcrumbs, expandable FAQs, price and availability, recipe times, event dates: a result that takes up more space draws more eyes, and therefore more clicks at the same position.
  • Entity understanding. Organization markup with sameAs (your social profiles, your Google listing) helps engines tie all your presences to one business rather than to scattered signals.
  • Disambiguation. Two brands with similar names, a keyword with several meanings, an author sharing a name with someone else: structured data settles it.

Why it matters even more for AI answers

Answer engines (ChatGPT, Perplexity, Gemini, Google’s AI overviews) work by extraction and synthesis. In a matter of seconds they must decide which claim on a page is reliable and attributable. A page whose facts are already declared in machine-readable form makes that job far easier.

Three effects worth remembering:

  • Your facts are explicit. Company name, service area, offerings, pricing, contact details: no longer inferred from a marketing paragraph.
  • Attribution is safer. A properly marked-up Article (author, publisher, dates) gives a model what it needs to cite you by name instead of paraphrasing you sourcelessly.
  • The Q&A format lines up naturally. An FAQPage already supplies short, self-contained question/answer pairs — exactly the shape an answer engine is looking for.

No model promises to cite marked-up sites more often. But with equivalent content, the page that made its information readable starts with a structural advantage.

The types to add first

There’s no need to mark up everything. Four types cover the vast majority of sites:

  1. Organization (or LocalBusiness) on the homepage: name, logo, URL, contact details, sameAs. For a shop or local service, LocalBusiness adds address, opening hours and service area — the backbone of local SEO.
  2. WebSite with the site name and, where relevant, the internal search action.
  3. Article / BlogPosting on every editorial page: headline, author, datePublished, dateModified, image.
  4. BreadcrumbList on deeper pages: breadcrumbs are frequently displayed as-is in results.

Depending on your business, add Product (with offers), Service, Event, Recipe or FAQPage next — only where the content genuinely matches.

The most common mistakes

  • Marking up information that isn’t on the page. Rule number one, and the most broken.
  • Self-assigned ratings. aggregateRating on your own organisation, with no verifiable reviews displayed, is explicitly discouraged.
  • Forgetting dateModified on updated content: you lose the freshness benefit.
  • Stacking contradictory blocks. Two JSON-LD scripts declaring different organisation names create more confusion than they resolve — a single coherent @graph is better.
  • Adding the markup and never checking it again. A redesign, a plugin change or a misconfigured cache can make it disappear with no visible difference on the page.

How to verify it works

Three reflexes, in order:

  1. Google’s Rich Results Test tells you whether your page is eligible for an enhanced display, and flags missing properties.
  2. The Schema.org validator checks the markup’s conformity itself, including types Google doesn’t use.
  3. Search Console’s “Rich results” report shows what Google actually detected across the whole site, with recurring errors — the only one of the three that reflects the real state of your indexed site.

Check after every redesign, every migration and every major CMS update.

Want to go further?

Digi-Pilot’s audit crawler detects the JSON-LD blocks present on each analysed page, identifies the declared types and flags editorial pages that have none — so you can spot in a single pass the content that’s invisible to search engines and AI. Start with a free SEO check of your site, then create an account to run a full audit and track your structured data over time.