Structured data that machines trust
Schema.org markup tells search engines and AI assistants who you are, what a page is about and which facts are reliable. We generate it from your content, link it into one graph and validate it on every deploy.
- JSON-LD, generated
- Validated in CI
- 1–3 week implementation
Two audiences, one markup
For classic search, structured data unlocks rich results: FAQs under your listing, breadcrumbs instead of URLs, event dates, product prices and review stars. Those features raise click-through on the same ranking.
For AI engines, the same markup does a different job. A retrieval system that reads Organization, Service and FAQPage blocks gets your facts without having to infer them from prose. When facts are explicit and consistent, the model is less likely to hallucinate a founding year or merge you with a similarly named company. This is why schema is on every GEO checklist.
The failure mode is familiar: markup pasted by hand into a template, forgotten, and now describing a product you no longer sell. We treat schema as code that is generated, tested and deployed.
{
"@context": "https://schema.org",
"@graph": [
{
"@type": "Organization",
"@id": "https://example.com/#organization",
"name": "Example Ltd",
"legalName": "Example Software Limited",
"url": "https://example.com/",
"foundingDate": "2006",
"sameAs": ["https://www.linkedin.com/company/example"]
},
{
"@type": "Service",
"@id": "https://example.com/services/audit/#page",
"name": "Technical SEO Audit",
"serviceType": "Technical SEO Audit",
"provider": { "@id": "https://example.com/#organization" },
"areaServed": "GB",
"url": "https://example.com/services/audit/"
},
{
"@type": "FAQPage",
"mainEntity": [{
"@type": "Question",
"name": "How long does an audit take?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Two to three weeks for most sites."
}
}]
}
]
}The schema types that earn their place
Schema.org defines hundreds of types. Fewer than ten do most of the work for a business site. We implement these first and add others only when a page genuinely represents that thing.
| Type | Use on | What it does for you |
|---|---|---|
Organization | Every page, once, with an @id | Establishes identity: name, legal name, founding date, locations, logo, sameAs profiles. The anchor for everything else. |
Service | Service pages | States what you offer, who provides it and where. Links to the Organization by @id. |
FAQPage | Pages with visible Q&A | Question-shaped facts for rich results and for AI retrieval. Must match on-page text. |
Article | Guides, insights, news | Author, publisher, published and modified dates. Freshness signals for engines that care. |
BreadcrumbList | Every inner page | Site structure for machines; breadcrumb display in results. |
Product | Product and pricing pages | Offers, price, availability, aggregate rating when shown. Essential for retail and e-commerce. |
Event | Event listings and ticket pages | Dates, venue, offers, status. Built into our registration and ticketing platform. |
DefinedTerm | Glossary entries | A named term with a definition inside a DefinedTermSet. Exactly how our glossary is marked up. |
LocalBusiness | Office and location pages | Address, geo, opening hours, phone per location. See local SEO. |
Generated, linked, validated
- Week 1
Inventory and mapping
Every template mapped to its types and the fields that feed them. Gaps in the content model are listed: no author field, no dates, no address per office.
- Week 1–2
Generator
A function in your CMS, framework or build that emits one
@graphper page, with stable@idvalues so the Organization, pages, breadcrumbs and FAQs reference each other. - Week 2
Validation in the pipeline
Schema validated on every build with the Schema.org validator rules and Google's rich result requirements. A broken graph fails the deploy, not a Monday morning report.
- Week 3
Monitor
Search Console enhancement reports watched for errors and warnings per type. AI engines checked with our visibility monitor for improved fact accuracy.
Mistakes we see most
- Markup for content that is not on the page
- A new Organization block on every page with no shared
@id - Review and rating markup without visible reviews
- FAQ markup on every page using the same five questions
- Hand-written JSON-LD that no longer matches the template
- Dates that never change, or that change on every build
- Plugins that emit three conflicting graphs
What good looks like
- One graph per page, one Organization with a stable
@id - Every field sourced from the content that renders the page
- Validation that fails the build
- Visible dates and authors that match the markup
Where schema fits in the wider work
Structured data, defined
The glossary entry to share with non-technical colleagues.
Read itllms.txt & AI crawlers
The other machine-readable file your site should have.
Set it upTechnical SEO Audit
Schema validation is one of seven audit areas.
Scope an auditKnowledge graphs
Why entities and relationships matter to search and AI.
Learn moreFrequently asked questions
Does structured data improve rankings?
Not directly. It makes pages eligible for rich results, helps search engines and AI engines understand entities and facts with less guesswork, and reduces the chance of being misread. The ranking benefit is indirect: better understanding, better click-through, fewer errors.
JSON-LD, Microdata or RDFa?
JSON-LD. It is the format search engines recommend, it is separate from your HTML so templates stay clean, and it can be generated from page metadata at build or request time. We only use Microdata when a legacy platform forces it.
Can we mark up content that is not visible on the page?
No. Markup must describe content the visitor can see. FAQ markup for questions that are not on the page, or review markup for reviews that are not shown, is against guidelines and can lead to a manual action.
How do you keep markup correct as the site changes?
By generating it from the same data that renders the page, and by validating it in the build or deployment pipeline. Hand-pasted JSON-LD drifts within months. This site generates all of its JSON-LD from page metadata at build time.
Which schema types matter for AI engines?
The ones that establish identity and facts: Organization with sameAs, Service or Product, FAQPage, Article with dates and authors, DefinedTerm for glossaries and BreadcrumbList for site structure. They give an engine the same facts in a form it does not have to infer.