Welcome, Product Hunt! Limen crosses its own threshold today. Your access code to everything premium is waiting in the launch thread.

Get your code
limen
Sign inCheck your site
I have watched people announce themselves at a thousand doors. Most trust the room to work out who they are. Schema is the visitor who states his name plainly at the threshold — and is remembered correctly. The machine no longer guesses at you. It simply reads the card you thought to hand it.
— Janus, at the door
Glossary
Glossary

JSON-LD Schema

JSON-LD schema is a way of writing down the facts on a web page in a form a machine can read without guessing. The letters stand for JavaScript Object Notation for Linked Data, but the idea underneath is simpler than the name. A human reading your page sees a business name in a nice font, an address in the footer, a price beside a product. A machine sees a wall of text and has to infer which part is which. Schema removes the inference: it is a quiet, invisible label tucked into the page's code that says, in effect, this is the business name, this is the address, this is the price — stated plainly, in a vocabulary that machines already agree on.

The shared vocabulary matters. Schema.org, a standard maintained across the major search and AI companies, defines the words — LocalBusiness, Product, Person, Review, and hundreds more — so that everyone's machines mean the same thing by them. JSON-LD is simply the tidiest way to write that vocabulary down: a self-contained block that sits in the page's head, apart from the visible content, doing its labelling without disturbing the design. Because it is separate, it doesn't change how the page looks to a person at all. It speaks only to the machines.

Where it goes wrong is usually one of two ways. Either it is absent — the page states its facts only in prose, and the machine is left to work them out, which it does imperfectly — or it is present but mismatched, claiming a price or a name that no longer matches what the page actually shows. A stale or contradictory label is worse than none, because it invites a machine to repeat something untrue with confidence. Good schema is accurate, current, and modest: it labels what is genuinely on the page, and nothing it can't back up.

The human translation is this. Imagine handing someone your business card versus making them piece your details together from an overheard conversation. Both can work, but the card is faster, cleaner, and far less likely to be misremembered. Schema is the business card you hand to the machines that now decide whether to recommend you. You are not writing it for a person to read — you are making sure that when an assistant is asked who does this, where, for whom, the answer about you is already spelled out, and spelled out right.