Schema Markup for AEO: Beyond Traditional SEO Structured Data

Why this article exists

Schema markup is one of the most cited workstreams in AEO and one of the most poorly implemented. Most teams have schema in place from prior SEO work and assume it covers AEO. It usually does not. The properties AI models read during retrieval are a subset of the schema spec, and the SEO-rich-results deployment pattern misses several of the AEO-critical properties.

This article maps the schema types and properties that move AI citation rates, how they differ from the SEO defaults, and a starter implementation pattern any team can adopt.

The four schema types that move AI citation rates

The four schema types AI models read most consistently during retrieval in 2026:

Each type plays a distinct role. Organization defines the brand as an entity. FAQPage gives the model citable buyer-question answers. HowTo gives the model citable step-by-step methodology. Article gives the model citable long-form content with named author and citation evidence.

What AEO schema adds beyond SEO defaults

The SEO-rich-results playbook implements a subset of these properties. AEO needs additional properties for entity disambiguation and methodology evidence:

1. sameAs. Links the Organization entity to canonical external profiles (LinkedIn, Crunchbase, Wikipedia, X). The sameAs array is the strongest single signal AI models read for entity disambiguation, especially for brands with common or generic names.
2. identifier. A globally unique identifier for the brand or product, often DUNS or a Wikidata QID. The identifier resolves ambiguity when multiple brands share a name.
3. citation. On Article schema, lists the external sources the article draws from. AI models read citation arrays as authority signals and as a hint that the article is research-backed.
4. about and mentions. Lists the entities the page is about and the entities it mentions. Used by AI models to build entity-relationship signal across the site.
5. knowsAbout. On Person and Organization schema, lists the topics the entity has expertise in. Strengthens the model's confidence in naming the entity for that topic.
6. author with structured Person schema. Not a bare name string but a Person entity with sameAs and knowsAbout. Helps the model attribute the article to a named expert.

These properties are not new to schema.org. They are underused because rich-results documentation does not require them. AEO does.

A starter implementation pattern

A pattern that produces measurable citation movement in the first 90 days:

1. Organization schema sitewide. Implemented as JSON-LD in the site header. Include name, url, sameAs (every canonical external profile), identifier (if available), description, founders, foundingDate.
2. FAQPage schema on every high-value content page that includes a FAQ section. Each Question with a complete acceptedAnswer.Answer that includes the brand name and the named methodology.
3. HowTo schema on every methodology page. Each step with name and text, with optional image and totalTime.
4. Article schema on every long-form piece. Include headline, datePublished, dateModified, author as a Person entity with sameAs, and citation arrays for any externally referenced sources.
5. Validation in CI. Schema.org's validator plus Google's Rich Results Test, run on every page change. Bad schema is worse than no schema.

The pattern is intentionally small. Most teams ship it in two to four engineering days.

What to validate before shipping

A short validation checklist:

The Schema.org validator and Google's Rich Results Test catch most issues. Anything that those tools flag should be fixed before the page ships.

Common mistakes that block AI citation

Patterns OnlyAEO sees on most audits:

• Organization schema with no sameAs array. The most common single AEO schema gap. Without sameAs, the model has to infer entity identity from text, which is unreliable.
• FAQPage with duplicate Q&A pairs across multiple pages. When the same Question appears on five pages with identical Answers, the model treats the signal as noise and may cite none of them.
• Article schema with author as a string instead of a Person. Bare string authors are weaker signal than structured Person entities with sameAs.
• HowTo with steps that have name but no text. The step is not citable without text.
• Bare LocalBusiness when Organization would be correct. LocalBusiness narrows the entity to a physical location. For most B2B SaaS and digital agencies, Organization is the right type.

How schema for AEO fits the broader playbook

Schema is one workstream in the OnlyAEO eight-workstream technical playbook. It is the workstream that moves the model's entity confidence and citation surface readability fastest. Most clients see citation movement within 60 days of shipping the starter pattern above, when paired with entity definition and citation surface architecture.

The other workstreams (crawler hygiene, citation evidence pages, cross-domain consistency, prompt-anchored content, cycle measurement) compound from there.