AI search is changing how people discover information.
Google has AI Overviews. ChatGPT can search the web. Perplexity answers questions using web sources. Other AI systems are also becoming interfaces for finding products, companies, services, and information.
That made me curious about something:
How well are normal websites actually prepared to be understood by AI systems?
Not just from a traditional SEO perspective.
I mean:
Can a machine clearly understand what this website is, who is behind it, what it offers, and how the information on the site is connected?
So I started looking at websites through that lens.
And one thing became obvious:
Having a website that is readable to humans doesn't necessarily mean it is equally understandable to machines.
Schema markup is only part of the picture
Most developers who have worked with technical SEO have probably seen JSON-LD.
For example:
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Example Company",
"url": "https://example.com/"
}
That's useful.
But imagine a machine trying to understand this company.
It may still have questions:
What exactly does this organization do?
Which website belongs to it?
Which social profiles represent the same organization?
What products or services does it provide?
Who are the people associated with it?
What other entities is it connected to?
Are different pages on the site talking about the same organization?
Simply declaring:
"@type": "Organization"
doesn't answer all of those questions.
The entity is more important than the markup itself
Consider a fictional company:
Acme Digital
Its homepage might say:
Acme Digital is a web development agency specializing in WordPress and ecommerce.
A human can understand that immediately.
But we can provide machines with much more explicit context through structured data.
For example:
{
"@context": "https://schema.org",
"@type": "Organization",
"@id": "https://example.com/#organization",
"name": "Acme Digital",
"url": "https://example.com/",
"sameAs": [
"https://www.linkedin.com/company/acme-digital/"
],
"knowsAbout": [
"WordPress",
"Ecommerce",
"Web development"
]
}
The important part isn't that every website needs exactly this structure.
The important part is the relationships and identity signals.
We're no longer just saying:
"This is an organization."
We're providing additional context about which organization we're talking about and what it represents.
But here's where things get interesting
A website can have schema markup and still have a weak implementation.
I've seen patterns such as:
Schema that exists but doesn't accurately represent the page.
Multiple plugins generating overlapping structured data.
Different pages using inconsistent organization information.
Missing or poorly implemented entity identifiers.
Missing identity references.
Important relationships between entities not being expressed.
Schema generated automatically without much thought about the underlying entity model.
So there is an important distinction:
A website having schema is not the same as a website having useful structured data.
Valid JSON-LD is only one part of the problem.
I started thinking about "AI readiness"
Instead of asking one question:
"Does this website have schema?"
I wanted to look at several signals together.
For example:
- Entity identification
Can we clearly identify the primary entity represented by the website?
That might be a:
company
person
product
organization
local business
publication
- Structured data
Does the website provide meaningful Schema.org information?
And does that information actually correspond to the content?
- Entity relationships
Are important relationships represented?
For example:
Organization
↓
Website
↓
WebPage
↓
Product / Service
↓
Person
The exact relationships depend on the website, but the concept is important.
- Identity signals
Can different sources on the web be connected to the same entity?
This is where properties such as sameAs can become useful.
- Consistency
Does the organization have the same name, URL, description and other important information throughout the website?
- Content understanding
Can a machine easily determine what a particular page is about?
These aren't a magic formula for getting into ChatGPT or Google AI results.
That's not what I'm claiming.
They're simply different pieces of the larger machine-understanding problem.
Why should developers care?
I think this goes beyond SEO.
We're increasingly building websites that aren't only consumed by humans.
They're consumed by:
search engines
crawlers
knowledge graphs
AI systems
recommendation systems
automated agents
other software
For years, we've optimized websites primarily around the human experience.
That's obviously still important.
But we're entering a world where machine understanding is becoming another important consideration.
If a website represents a business, product, person, or organization, we should make that information as clear and consistent for machines as we do for humans.
I built a free checker while working on this
This led me to build AI Schema Gen.
It started as a schema-focused project, but I gradually expanded it into an AI-readiness checker.
You can enter a website and get an assessment of things such as:
structured data
schema implementation
entity information
identity signals
relationships
other AI-readiness signals
You can try it here:
https://www.aischemagen.com/ai-readiness-checker/
The scanner is free.
I'm especially interested in seeing what developers discover when they run their own websites through it.
One thing I don't believe
I don't think there's a magic Schema.org type that will suddenly make your website appear in ChatGPT.
I also don't think adding dozens of schema types automatically makes a website "more AI-ready."
In fact, inaccurate, irrelevant, duplicated or misleading structured data can create more problems than it solves.
The goal should be:
Accurate information + clear entities + meaningful relationships + consistent implementation.
Not:
"Add as much schema as possible."
That's the direction I'm exploring with AI Schema Gen.
I'd genuinely like to hear from other developers:
What do you think makes a website "AI-ready"?
And if you run your website through the checker, let me know what it finds.
Top comments (0)