There has been a lot of discussion in the SEO community about whether Schema markup influences visibility in Google’s AI Mode.

Some suggest that having certain Schema types in place could increase the chances of being cited. To test this theory, we analysed 107,352 websites that appeared as citations in AI Mode to see whether Schema markup had any direct impact.

Our findings show that while Schema is widely present, no single type beyond the standard set (Organization, WebPage, Article, etc) appears to give a measurable advantage.

The role of Schema in AI Mode

Schema markup has long been part of the SEO toolkit for enhancing search visibility. From featured snippets to rich results, structured data provides context that helps search engines understand the meaning of content.

With the rise of AI Mode in Google Search, a number of have SEOs speculated that specific Schema types such as FAQs, HowTo or Product might drive inclusion as a cited source.

Our research shows that this might not be the case.

What the data shows

When we analysed Schema adoption across the 107,352 URLs cited in AI Mode, the distribution aligned closely with the Schemas you would expect on any well-structured website:

Schema Type
% of AI Overviews containing it
Organisation 82%
WebPage / Article 76%
BreadcrumbList 59%
FAQPage / QAPage 41%
Person (Author) 38%
Product / Service 34%
ImageObject 28%
Review / AggregateRating 19%

The data suggests that Google is not favouring special Schemas to select citations in AI Mode.

Instead, the baseline structures such as Organisation, Article, Breadcrumbs and Author are the most common. These are the same Schemas that form the backbone of most content sites.

Common Schema triplets

We also looked at the most frequent Schema connections (triplets) found in AI Mode citations:

  • WebPage → mainEntity → Article
  • Article → author → Person
  • Organisation → url → Homepage

These represent very standard relationships that describe ownership, authorship, and the connection between a page and its content. Nothing out of the ordinary emerged.

Why this makes sense

It is not surprising that Schema does not appear to influence citation in AI Mode. While Google’s AI Mode is built on top of Google Search infrastructure (which does process structured data), the way large language models (LLMs) interpret web pages is different.

Other LLMs, such as Claude or ChatGPT, do not read Schema as a neat JSON block. Instead, they tokenise the raw HTML of a page. Tokenisation is the process of breaking text down into smaller units (tokens) that the model can process.

For example:
{
"@type": "Organisation",
"name": "Example Corp",
"url": "https://example.com"
}

To a human, that is a clear, structured statement that “Example Corp” is the organisation with a homepage at example.com. To an LLM, it gets split into tokens such as:


{, "@type", ":", "Organisation", ",", "name", ":", "Example", "Corp", ...

The model does not see it as a structured Schema graph. It just sees fragments of text and punctuation, the same way it would see body copy or inline code. The semantic relationships (for example, “Organisation → name → Example Corp”) are flattened.

This means Schema does not provide the same advantage in an LLM-driven environment that it does in traditional search. While Google’s ranking systems may use Schema to better understand and classify pages, the AI layer is drawing from tokenised text rather than structured nodes.

In practice:

  • Search infrastructure can parse JSON-LD and RDFa as structured data.
  • LLMs reduce everything, including Schema and HTML, into a stream of tokens.

This helps explain why Schema is unlikely to be a deciding factor for inclusion in AI Mode. It gives context to Search, but the LLM component processes it like any other text.

What this means for SEOs

Schema markup still matters. It gives Google structured context and improves eligibility for rich results. But in the context of AI Mode, it does not appear that layering on niche Schema types will give you an edge.

Instead, focus on:

  • Accuracy: Ensure your Schema reflects reality and is not misleading.
  • Completeness: Implement the fundamentals consistently across your site.
  • Content quality: Remember that citations in AI Mode are ultimately driven by content authority and relevance, not markup tricks.

Conclusion

Our test shows that Schema is a hygiene factor (at best) for AI Mode visibility, not a differentiator.

Having the basics in place (Organisation, Article, Person, Breadcrumbs) is important, but no evidence suggests that Schema alone determines whether you surface as a citation.

Put simply, Schema helps Google understand your content, but it will not guarantee you a spot in AI Mode.