Article overview
When a search engine like Google or an AI such as ChatGPT visits your site, it may have difficulty understanding who you are, who you are for and what you sell.
JSON-LD is machine-readable code that sits inside your website and helps solve that problem. But its value goes beyond site-level clarity.
As part of a Generative Engine Optimisation (GEO) strategy, its most valuable use is building a clear and consistent identity around the business itself, making it easier for AI systems to identify, assess and recommend.
In short, JSON-LD helps you become more visible to AI and makes it easier for AI to confidently recommend you to the right buyers.
What is a JSON-LD file and why does it matter for your business?
What’s clear to humans is often a muddle to AI.
Your website may explain who you are, what you do and who you help. But it’s written for people, and people are very good at inferring what you mean and filling in gaps.
Search engines and AI systems are less reliable.
They may not clearly understand:
which business the website represents;
whether your personal name and company name refer to the same entity;
which services or products belong to your business;
which people or organisations are connected to it;
whether external profiles, such as LinkedIn, refer to the same business;
how your pages relate to one another.
That might not seem important, but if visiting AIs and AI-powered search engines such as Google or Bing can’t determine all this quickly and easily, they’re more likely to skip over you in favour of a business they can understand with greater confidence.
AI visibility begins with clarity, and JSON-LD provides that clarity.
JSON-LD files help search engines and AI systems by presenting key information about your business in a way AI can easily digest.
For an Irish business, this matters because more of your buyers are using AI to find, compare and assess potential suppliers. Gartner predicts that brands’ organic search traffic will fall by 50% or more by 2028 as consumers increasingly use generative AI-powered search.
The better you are at helping AI understand your value and relevance, the better your chances of being the business it recommends.
But not all JSON-LD files are created equal.
One approach to writing them, and by far the most common, is the AI-SEO approach. The other, and far more valuable, is the Entity-GEO, or Generative Engine Optimisation, approach.
What’s the difference between AI-SEO and Entity-GEO?
AI-SEO isn’t the same as Generative Engine Optimisation (GEO). That difference matters. A lot.
AI-SEO uses JSON-LD to describe individual pages on your site. This creates clarity around what you’re saying.
Entity-GEO uses JSON-LD to define the business behind those pages and connect its people, products, services and supporting information. That creates clarity around who you are.
AI will be more confident when deciding whether to recommend you if it’s clear about who you or your business are. That identifiable person or business is called an ‘entity’.
When writing your JSON-LD, prioritise building and clarifying that entity.
What does JSON-LD look like?
A simple JSON-LD block might look like this:
Your visitors won’t see this. JSON-LD is hidden. But search engines and AIs can see it.
It tells these machines what the business is, what it does, who it serves, where it operates and any other information that helps the AI understand it.
The sample above is just a snippet to give you an idea of what it looks like. A full JSON-LD implementation will usually contain more information and more connections.
Depending on the business, that may include:
the official business name;
the website connected to it;
the people who own, run or represent it;
the products or services it provides;
its brands or product ranges;
the locations or markets it serves;
external profiles that confirm its identity;
the relationships between the business, its people and its offers.
An Irish manufacturer may need to connect products, brands, distributors and technical information back to the same company.
A solicitor’s practice may need to connect the firm, its solicitors, its specialist services and the locations it serves.
A high-value retailer may need to connect products, brands, showrooms, warranties and purchasing information.
There isn’t one standard template that works equally well for every business.
The JSON-LD should reflect what the business actually is.
Can you use AI to write JSON-LD?
Yes, AI can help you write JSON-LD, but you still need enough expertise to judge what it produces.
AI is very good at generating code that looks plausible, but that doesn’t mean the code is accurate, complete or strategically useful.
It may choose the wrong schema type, create weak connections, invent properties or produce markup that describes individual pages without properly defining the business behind them.
If you don’t have the expertise to know whether the output is good, hand the work to someone who does.
That expertise rule should apply to almost everything you do with AI.
Do you really need JSON-LD on your site?
If you want AI to recommend or favourably assess your business, then yes, you should treat JSON-LD as a basic part of your website infrastructure.
Without it, you’re asking search engines and AI systems to work out your identity, products, services and relationships from your written content alone.
They may succeed, but they may also misunderstand you, fail to connect important information or choose a competitor whose business is easier to interpret.
JSON-LD alone doesn’t guarantee that AI will recommend you. No single file or technical fix can do that.
What it does is improve the conditions under which a recommendation can happen.
Does that mean every business needs JSON-LD?
There are still businesses without a website and businesses using off-the-shelf Gmail addresses for commercial email.
For businesses like that, the answer is clearly no.
For the rest of us, clearer machine understanding is an opportunity with considerable potential upside.
________
If you’d like to talk to me about setting up your business to be seen and recommended by AI, while also having human-focused copy that your potential buyers will respond positively to, click here to email me.
Sources
Sources
Google Search Central, Introduction to structured data
https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data
Google Search Central, Organisation structured data
https://developers.google.com/search/docs/appearance/structured-data/organization
Google Search Central, AI features and your website
https://developers.google.com/search/docs/appearance/ai-features
Gartner, Gartner Predicts 50% of Consumers Will Significantly Limit Their Interactions with Social Media by 2025
https://www.gartner.com/en/newsroom/press-releases/2023-12-14-gartner-predicts-fifty-percent-of-consumers-will-significantly-limit-their-interactions-with-social-media-by-2025
Schema.org
https://schema.org/
Schema.org, sameAs
https://schema.org/sameAs
Schema.org, Service
https://schema.org/Service
W3C, JSON-LD 1.1
https://www.w3.org/TR/json-ld11/
OpenAI, Publishers and developers FAQ
https://help.openai.com/en/articles/12627856-publishers-and-developers-faq
Details
Category
GEO
Reading
6 minutes
Author

Paul Melrose
Paul Melrose is a Dublin-based, Irish copywriter and Generative-Engine-Optimisation specialist.
More articles




