5LLMs monitored
6perception metrics
30+Academy lessons
EUAI Act ready
Back to Guides
AdvancedGEO Optimization

Structured Data for LLMs

Technical guide to JSON-LD schemas, llms.txt files, and other markup that helps AI understand your content.

15 min readFebruary 10, 2024

Why structured data matters for AI

Unstructured content requires AI to infer relationships. Structured data makes relationships explicit.

When AI encounters "Acme was founded in 2019", it must determine that Acme is a company and 2019 is a founding date.

With Schema.org markup, you state explicitly: Organization, foundingDate: 2019. No inference needed.

Structured data doesn't guarantee AI will use your information, but it dramatically reduces misinterpretation.

Essential Schema.org types

Organization - Your company entity. Include: name, foundingDate, founders, numberOfEmployees, address, sameAs (social profiles)

Product or SoftwareApplication - Your products. Include: name, description, offers (pricing), aggregateRating, featureList

FAQPage - Question and answer content. Highly valuable for AI retrieval.

HowTo - Step-by-step guides. AI can extract and cite these directly.

Review - Customer feedback with ratings. Contributes to sentiment perception.

Person - Founders, executives, key team members. Helps with entity disambiguation.

Implementation basics

Use JSON-LD format in your <head> tag:

<script type="application/ld+json">

{"@context": "https://schema.org", "@type": "Organization", "name": "Acme Software", "foundingDate": "2019", "numberOfEmployees": "150"}

</script>

Every page should have relevant schema. Product pages need Product schema. About pages need Organization schema. FAQ pages need FAQPage schema.

Test with Google's Rich Results Test and Schema.org validator.

The llms.txt standard

llms.txt is an emerging standard for communicating with AI systems directly. Think robots.txt but for language models.

Place at /llms.txt on your domain. Include:

# Brand Information

Name: Acme Software

Description: B2B inventory management platform

Founded: 2019

Headquarters: Berlin, Germany

# Key Facts

- Serves 10,000+ businesses

- Available in 12 languages

- SOC 2 Type II certified

While not universally adopted yet, it's a low-effort addition that helps forward-thinking AI systems.

Advanced: Entity linking

Connect your entities to authoritative databases:

sameAs: Link to your Wikipedia page, Crunchbase profile, LinkedIn company page

identifier: Include stock tickers, DUNS numbers, or other business identifiers

@id: Use consistent URIs for entities across your site

Entity linking helps AI understand you're the same "Acme" mentioned in other sources, not a different company.

Common mistakes to avoid

1. Incomplete schema - Partial data is worse than no data. If you add Product schema, include all relevant fields.

2. Inconsistent across pages - Your Organization schema should be identical everywhere it appears.

3. Outdated information - Schema with old pricing or discontinued features causes accuracy issues.

4. Over-optimization - Don't stuff schema with keywords. AI systems can detect this.

5. Missing on key pages - Homepage, pricing, and about pages absolutely need structured data.

Ready to apply what you learned?

Run a free perception audit and see how these concepts apply to your brand.

Start Free Audit