JSON-LD
JSON-LD (JavaScript Object Notation for Linked Data) is a way to embed structured data in a web page as a single script block, separate from the visible text. It labels facts about the page — that this is an Organization, a Product, an Event, an FAQ — using the Schema.org vocabulary, so a crawler doesn’t have to infer meaning from prose. It is the most common syntax for structured data on the modern web, alongside older formats like Microdata and RDFa.
Why JSON-LD matters for AI visibility
Structured data makes a page easier to parse correctly, but it is worth being precise about what it does and doesn’t do. Google’s own guidance on AI Overviews and AI Mode states plainly that no special markup, schema, or AI-specific file is required to appear in them, and explicitly warns against writing content "for AI" instead of for readers. JSON-LD is not a proven independent ranking or citation factor. What it does reliably do is remove ambiguity: it tells a system unambiguously that a name is a business name, that a number is a price, that a date is an opening hour. That clarity can help a page get parsed and represented correctly once an AI system or search engine has decided to look at it — it does not substitute for the content, mentions, and citations that get you noticed in the first place.
How JSON-LD works technically
JSON-LD usually sits as its own code block in the head area or at the end of the HTML page, wrapped in a <script type="application/ld+json"> tag. Unlike Microdata, it does not weave attributes into the visible markup; it lies separately alongside it, which keeps your layout untouched and the data easier to maintain or generate programmatically. Content-wise, it follows the Schema.org vocabulary: you declare a @type (Organization, Product, FAQPage, Event, and so on) and fill in the properties that type defines — name, address, price, datePublished. Search engine crawlers and AI crawlers parse this block directly and can map the values to entities they already track.
Common mistakes
The classic mistake is markup that disagrees with the visible page: if your JSON-LD lists a price of €89 but the page shows €129, that mismatch reads as manipulation, not helpfulness. Just as common are missing required fields, badly formatted dates, or a @type that doesn’t fit the content. Stale data causes similar damage — opening hours nobody updates, an event date that already passed. A single misplaced comma can also invalidate the entire block, so run it through a validator (Google’s Rich Results Test or the Schema.org validator) before publishing, and keep it as current as the page itself. A subtler mistake is treating JSON-LD as a growth lever on its own: it is markup, not marketing, and it won’t compensate for thin content or a brand nobody mentions elsewhere.
Relation to structured data and GEO
JSON-LD is the implementation, structured data is the broader idea, and both sit inside generative engine optimization as supporting infrastructure rather than the main lever. The evidence so far points elsewhere for citation gains: an Ahrefs analysis of roughly 75,000 brands found that how often a brand is mentioned across the web correlates with AI citation rate at about 0.664 — roughly three times stronger than the correlation for backlinks, at about 0.218. FAQPage or Article schema can still help a system read a question-answer pair or byline correctly once it's looking at your page, which is a real, if modest, contribution to citability. Treat JSON-LD as good hygiene that removes friction, not as the mechanism that gets you cited.
Example
Imagine a small bakery in Regensburg. Its home page carries a JSON-LD block marking it up as a LocalBusiness: name, address, phone number, opening hours, and a note that it stocks gluten-free bread. If someone asks an AI assistant "where can I get gluten-free bread in Regensburg on a Sunday," a system that has already decided to draw on this page can read the hours and the product detail straight from the markup rather than guessing from a paragraph of prose. The JSON-LD doesn’t get the bakery found — mentions, reviews, and content do that — but it reduces the odds of the AI getting the hours or the detail wrong once it's there.
Common questions
Do I need programming skills for JSON-LD?
Basic knowledge helps but isn’t required. Most content management systems and SEO plugins generate JSON-LD automatically from your existing fields. What matters more is keeping the values accurate and current, and validating the block before it goes live.
Is JSON-LD the same as Schema.org?
No. Schema.org is the vocabulary — the set of types and properties like Organization, Product, or FAQPage. JSON-LD is one syntax for writing that vocabulary into a page (Microdata and RDFa are others). They work together but aren’t interchangeable terms.