Schema Markup for AEO: 10 Advanced JSON-LD Types You Are Not Using

Structured data has settled into a rut. Most sites ship Article, FAQPage, Organization and stop. That covers the basics and leaves a long list of types that describe your content far more precisely, several of which map directly onto the way answer engines retrieve and attribute information.

Which schema types help most with answer engines?

The ones that describe structure and provenance rather than just topic. HowTo, Dataset, ClaimReview, Speakable, DefinedTerm and QAPage tell an engine what kind of information a passage contains and where it came from, which is exactly what a retrieval system needs to decide whether to quote you.

The ten worth implementing

1. HowTo

Marks a genuine step sequence with tools, supplies and per-step detail. It maps cleanly onto procedural questions, which are a large share of what people ask assistants.

2. Dataset

Wildly under-used outside academia. If you publish original research, benchmarks or survey results, Dataset describes the data itself: what it measures, how it was collected, the licence and the format. It is one of the strongest provenance signals available.

3. ClaimReview

Built for fact-checking, and useful for any brand correcting a persistent misconception in its category. It states a claim, an assessment and evidence in a machine-readable form.

4. Speakable

Nominates the passages best suited to being read aloud. It is a direct instruction about which sentences carry your answer, which is useful well beyond voice.

5. DefinedTerm and DefinedTermSet

Turns a glossary into a structured vocabulary. If you own terminology in your category, this is how you attach yourself to those terms as an entity rather than a page.

6. QAPage

Distinct from FAQPage: it marks a single question with community answers, including accepted answers and vote counts. Correct for forum and support-thread structures where FAQPage would be a misuse.

7. SoftwareApplication

Describes an application with operating system, category, pricing and rating. Essential for tools and apps, and frequently missing.

8. Course and LearningResource

Describes educational content with level, prerequisites and time required. If you run any kind of academy or documentation, this is how it gets understood as learning material.

9. Review and AggregateRating done properly

Widely implemented and just as widely implemented wrongly. It must reflect genuine reviews, on the item being reviewed, without self-serving markup on your own products. Done correctly it is a strong trust signal, done incorrectly it earns a manual action.

10. Event

Under-used by service businesses running webinars and workshops. Real dates, locations, attendance modes and offers make an event discoverable in ways a landing page alone is not.

Rules that keep you safe

• Only mark up what is visible on the page. Invisible markup is a violation and models increasingly cross-check.
• Use one JSON-LD block with an @graph rather than several disconnected blocks.
• Connect entities with @id references so the engine reads one coherent description instead of fragments.
• Validate every deployment. A single malformed property can invalidate an entire block.
• Accuracy over coverage. Wrong structured data is worse than none.

Where to start

Pick the type that matches content you already have. If you publish research, add Dataset. If you have a glossary, add DefinedTerm. If you run webinars, add Event. Implementing one type properly beats sprinkling five types thinly across templates.

A free audit validates your existing markup, and the Academy has the full structured data chapter.