Schema markup is structured data that is written in a format called JSON-LD. It tells an engine what a page means, for example that this is a product, that this is its price, and that this is a review. When the markup is relevant and valid, it can make a page eligible for Google search features and easier for AI engines to read. Being eligible does not guarantee that the feature will actually show, and not every type is worth the effort.
What structured data can earn on the results page: a rating snippet and an FAQ dropdown. More of the result is yours, so more of the click is too.
The types that still pull their weight
- Organization and WebSite. The foundation. They tell engines who you are as an entity, which underpins both knowledge panels and how AI engines identify you.
- Product and Offer. For anything sold: price, availability and reviews can all surface in the result.
- FAQPage. This is a clear question-and-answer structure that AI engines can lift and that readers can scan easily. Google retired FAQ rich results in May 2026 (Google Search updates), so you should treat this as structure for people and for AI answers, and not as a Google rich result. This is covered in how to make your site answer-ready.
- Article and BreadcrumbList. Author, date and site structure, which feed the trust and navigation signals engines lean on.
- LocalBusiness. For anywhere with a physical presence, the backbone of local and map results.
Where teams get it wrong
A common schema mistake is markup that is not valid, which happens when a declared type is missing a property that is required, or when the JSON-LD will not parse. An engine cannot use syntax that is broken, so the page becomes eligible for none of the features that the markup was meant to support. A separate problem is markup that misrepresents the page, for example when you mark up content that is not visible. That can violate Google's structured data policies (sd-policies).
Trackers fire before anyone consents
Does schema help AI Visibility?
Schema does not make an engine cite a page, but it can make the page less ambiguous. An FAQ or product that is clearly marked up is easier for a model to read and to attribute correctly. When engines build answers out of passages, being easy for a machine to read is an advantage.
Schema markup FAQ
Which schema types are eligible for rich results?
Product, Review, Recipe, Event and Breadcrumb are some of the types that are eligible for visible search features, as long as they meet each feature's current requirements (Google search gallery). Google retired FAQ rich results in May 2026, so FAQPage no longer produces a Google rich result, although it still helps readers and AI answers. Organization and WebSite matter for helping engines understand your entity, even when there is no visible snippet.
Can too much schema hurt?
Marking up content that is not visible, or declaring types that the page does not really have, can violate Google's structured data policies (sd-policies). So you should mark up only what is genuinely on the page, and then validate it. Do not add types just to raise an audit count.
How do I check my structured data?
A schema.org validator checks the syntax, and Google's Rich Results Test checks whether a URL is eligible for a specific feature. A site-wide audit like Auditaar validates the structured data across the whole crawl and flags any type that is missing a required property. You should keep validity, eligibility and actual display separate from each other, because a valid result is not a guarantee that the feature will show.
Auditaar validates structured data across your whole site, flags invalid and incomplete markup, and scores it into the findability pillar. Run an audit, or read the top technical SEO mistakes.