Schema markup that actually helps AI search
Not all schema is equal. Here are the types that move the needle on AI Overviews, ChatGPT, Perplexity and Gemini — and the ones to skip.
Schema is not a ranking factor in the classic sense, but it is an eligibility factor. Without it, your content cannot be parsed into the structured candidate sets AI engines pull from.
Article
Required on every editorial page. Include headline, datePublished, dateModified, author and image. Most engines use dateModified as a freshness signal.
FAQPage
Pair every long-form post with 4–6 Q/A entries. FAQPage entries are extracted directly by AI Overviews and quoted by LLMs.
HowTo
For step-based tutorials only. Misused HowTo schema is a quick way to get a manual action — keep it real and granular.
Organization
Place it once on your home page. Include sameAs links to LinkedIn, Wikidata, Crunchbase. This is the backbone of your entity profile.
Schema you can skip
- BreadcrumbList helps SERPs but rarely changes LLM behavior.
- Speakable is barely used in 2026.
- Custom types — unless you have a strong reason, stick to the official vocabulary.
Frequently asked questions
- Does schema replace good HTML?
- No. It augments it. If your HTML is messy, schema will not save you.
- Where do I validate?
- Google Rich Results Test and Schema.org's validator. Run both.
- Does schema work for non-Google engines?
- Yes. ChatGPT, Perplexity and Gemini all parse JSON-LD.