Schema Markup for AEO: What Actually Works in 2026
Does schema markup improve AI citations? The honest answer is more nuanced than the standard advice — here's what the controlled studies show and which schema types actually matter.
Article
Does Schema Actually Improve AI Citations?
The standard advice for two years has been: add JSON-LD, mark up your FAQs, help the machines parse you. It sounded right — schema exists to make content machine-readable, and AI engines are machines reading content. Clean logic.
Then the controlled data came in.
Ahrefs ran a matched difference-in-differences study — 1,885 pages that added JSON-LD versus ~4,000 control pages that didn't, measuring citation changes across AI Overviews, AI Mode, and ChatGPT. The lift: roughly +2% — statistical noise. Meanwhile CapstonAI measured +90% average citation lift from FAQ schema in a different dataset. The studies genuinely disagree, which is why RankSenseAI calls schema the most contested factor in the entire body of research.
The honest take
Schema is not the citation lever the industry pretends it is — adding JSON-LD to already-cited pages won't meaningfully lift your rate. But it's also not theatre: it's entity infrastructure that makes your content correctly understood. Do it, just don't expect it to carry the page.
What Schema Actually Does
Entity disambiguation. Tells the engine whether your page is about Apple the company or apple the fruit. Doesn't earn the citation — makes sure you're correctly understood when retrieved. The entity SEO guide covers the full mechanism.
Rich result eligibility. The documented, boring, still-valuable benefit: stars, FAQs, pricing, recipes in classic Google results — which still drives most of your actual traffic.
Author-publisher entity edges. Person schema with sameAs ties content to real humans; Organization ties content to your brand. Authored content outperforms anonymous content across every engine we track — Erlin data puts structured facts behind an 8x coverage gap.
Freshness signals. datePublished + dateModified tell engines how current you are — and content under three months old is up to 3x more likely to be cited. A dateModified that actually updates is schema doing real work.
The Types That Matter
Not all schema is equal — the types that support AI citation declare entities and authorship, not rich-result bait.
| Type | What it does | Priority |
|---|---|---|
| Article/BlogPosting | Author, dates, publisher, mainEntityOfPage on every post | Required |
| FAQPage | Declares Q&A pairs engines extract directly | High — the +90% lift candidate |
| Organization | Your machine-readable identity card — name, url, logo, sameAs, founder | Required |
| Person | Ties posts to named humans with credentials | High — authored beats anonymous |
| HowTo | Step structure for process content | Situational |
| Product/Service | What you sell, priced offers | On offer pages |
Google deprecated FAQ rich results for most sites in 2024 — so FAQPage's value is now mostly AI extraction, not SERP decoration. Worth doing anyway.
The Five Properties Everyone Skips
Most implementations cover name/description/url and stop. The properties that actually matter:
- author → Person + sameAs — the entity edge to a real human
- mainEntityOfPage → @id — declares the canonical entity the page is about
- datePublished + dateModified — the freshness signal, kept honest
- publisher → Organization — connects content to your brand entity
- articleSection + keywords — topical classification for retrieval
What Schema Cannot Do
- Substitute for ranking. 76.1% of AI Overview citations come from top-10 pages. Schema wins ties among ranked pages; it doesn't get you ranked.
- Fix bad content. JSON-LD on thin, derivative content produces well-described thin, derivative content.
- Replace answer-first structure. The lead-answer pattern is the citation mechanic; schema is plumbing.
- Guarantee anything. Every honest study says the same thing — it's a foundation, not a tactic.
The Implementation
- Audit — Rich Results Test on your templates; find what's missing or invalid.
- Add the five properties above to every content page.
- Match type to page — Article on posts, FAQPage on Q&A sections, Organization on home/about, Service on offer pages.
- Validate — invalid schema is worse than none.
- Then forget it — the citation lift lives in answer-first structure, original data, and entity authority. Schema just makes sure the engines can read the good work. The AEO checklist scores it alongside the dimensions that actually move.
Frequently Asked Questions
Should I bother if it doesn't directly lift citations?
Yes — it's cheap entity infrastructure that supports everything else, plus rich-result eligibility for classic search. Just treat it as plumbing, not a growth lever.
Did Google really deprecate FAQ schema?
The rich results — for most sites, in 2024. FAQPage markup still helps AI engines parse Q&A structure, which is where its value lives now.
Schema or answer-first content — which first?
Answer-first, every time. Structure is the citation mechanic; schema is documentation for the structure.
JSON-LD or microdata?
JSON-LD — Google's recommended format, easier to maintain, most reliably parsed by AI crawlers.
Schema deployment is one of the automated checks in the Altyzo platform — $30K license for agencies. Or Altyzo Managed handles the whole stack.
For agencies
Take on more clients without hiring more people.
Altyzo is self-hosted AI that does the research, writing, publishing, and monitoring for your clients' SEO, AEO, and GEO. Your team reviews it. $30K lifetime license.