How Structured Data Influences LLM Brand Recommendations: An Early Analysis

Schema markup doesn't directly tell ChatGPT what to say. But it shapes the systems that AI tools rely on to understand your brand. Here's what we know, what we don't, and what the early evidence suggests.

There's a question circulating in the GEO community that doesn't have a clean answer yet: does structured data actually influence whether AI tools recommend your brand?

The honest answer is complicated. And the nuance matters, because getting it wrong leads to either wasted effort or missed opportunity.

Here's what we know so far.

The direct vs. indirect distinction

The most important thing to understand about structured data and LLMs is that the relationship is indirect, not direct. LLMs like ChatGPT, Claude, and Gemini don't crawl your website's schema markup the way Google's search crawler does. They don't read your JSON-LD and decide to recommend you based on what they find there.

But that doesn't mean structured data is irrelevant. Far from it.

LLMs build their understanding of the world through two channels: their training data (the massive corpus of web content they learned from) and their retrieval systems (the real-time web access that tools like Perplexity, ChatGPT with browsing, and Google's Gemini use to ground their responses in current information).

Structured data influences both channels, but indirectly. It shapes how search engines understand and categorize your content, which in turn shapes the search indexes and knowledge graphs that AI retrieval systems draw from. It makes your content easier for AI crawlers to parse during training data collection. And it reinforces the entity relationships that help AI tools connect your brand to the right topics and categories.

Think of it this way: structured data doesn't talk to the LLM. It talks to the systems that the LLM listens to.

What Microsoft has confirmed

Of all the major AI platform operators, Microsoft has been the most explicit about the role of structured data. Fabrice Canel, Principal Product Manager at Microsoft Bing, has stated publicly that schema markup helps Microsoft's LLMs understand web content. Given that Microsoft's Copilot is powered by the same infrastructure as Bing's search index, this is a direct confirmation that structured data feeds into at least one major AI platform's recommendation engine.

Google has been less direct but has signaled similar dynamics. Google's AI Overviews and AI Mode draw from Google's search infrastructure, which relies heavily on structured data for entity understanding and knowledge graph construction. When your Organization schema tells Google who you are and what you do, that information flows into the same systems that power Gemini's recommendations.

OpenAI and Anthropic haven't made public statements about structured data's role in their systems. But given that ChatGPT's browsing feature and Claude's web access both rely on crawling and parsing web pages, it's reasonable to infer that well-structured content, including pages with clear schema markup, is easier for these systems to process and extract useful information from.

The knowledge graph connection

This is where structured data's influence becomes most tangible.

Google's Knowledge Graph, Bing's entity index, and similar systems maintained by other platforms are massive databases that map relationships between entities: brands, people, products, concepts, locations. These knowledge graphs are a primary source of ground truth for AI systems. When an LLM needs to verify a fact or understand a relationship between entities, it often consults these structured databases.

Schema markup is one of the primary methods for getting your entities recognized in these knowledge graphs. Organization schema with proper sameAs links to Wikipedia, Wikidata, and authoritative profiles tells these systems exactly who you are and how you relate to other entities. Product schema tells them what you sell. Person schema for your team members establishes expertise associations.

Research from Data World found that LLMs grounded in knowledge graphs achieve significantly higher accuracy compared to those relying solely on unstructured text. This makes intuitive sense. An AI system that can verify "Company X makes Product Y in Category Z" against a structured knowledge graph is going to be more confident in that information than one piecing it together from scattered mentions across web pages.

The implication is clear: the more comprehensively and accurately your brand is represented in knowledge graphs, the more confidently AI systems can reference you in their responses. And structured data is the primary lever for influencing your knowledge graph presence.

What the skeptics get right

Not everyone in the SEO community is convinced that structured data matters for AI visibility, and some of their objections are valid.

Patrick Stox and Christopher Shin, both respected voices in the search community, have raised pointed questions about the logic chain connecting schema markup to LLM output. Their core argument: LLM training data comes from crawled web content, not from schema markup specifically. Search indexes are built from crawled data. So the path from "I added schema to my page" to "ChatGPT now recommends my brand" is neither direct nor guaranteed.

This critique is technically accurate. Adding Organization schema to your homepage will not, by itself, cause ChatGPT to start recommending your brand. Schema is not a ranking factor for LLMs any more than it's a direct ranking factor for Google (John Mueller confirmed in 2025 that structured data doesn't directly influence rankings).

But the skeptics sometimes overstate the case. The question isn't whether schema markup is a direct LLM ranking factor. It isn't. The question is whether it's part of a system of signals that influences how AI tools understand and reference your brand. And on that question, the evidence increasingly points to yes.

The schema types that matter most

If structured data's influence on AI visibility is indirect but real, which types of schema markup are most worth implementing? Based on the current evidence, five types stand out.

Organization schema is the foundation. It establishes your brand as a defined entity with specific attributes: name, description, URL, logo, social profiles, founding date, location. The sameAs property is particularly important because it links your entity to authoritative external references like Wikipedia and Wikidata, which knowledge graphs use for verification.

Article and BlogPosting schema signals key content attributes like publication date, author, publisher, and topic. These help AI systems assess freshness and authority when deciding which sources to draw from. Including dateModified is especially valuable because retrieval systems increasingly prioritize recently updated content.

FAQPage schema matches the natural question-and-answer format that AI tools use to generate responses. When an LLM is looking for a clear answer to a specific question, a page with structured FAQ markup that directly addresses that question is easier to parse and cite than one that buries the answer in narrative text.

Product and Service schema provides detailed, machine-readable information about what you offer: features, pricing, availability, reviews. For e-commerce and SaaS brands, this structured product information can feed directly into the comparison data that AI tools draw from when making purchase recommendations.

Person schema for key team members and content authors strengthens E-E-A-T signals by connecting your content to identifiable individuals with verifiable expertise. When an AI tool is assessing the authority of a source, being able to map the content to a real person with established credentials in the relevant field increases confidence.

A practical framework for implementation

Given that structured data's influence on AI visibility is indirect, how should brands prioritize their implementation?

Start with the highest-leverage items. If you don't have Organization schema on your homepage with proper sameAs links, that's step one. This is the single most impactful action for establishing your brand as a recognized entity in knowledge graphs.

Next, ensure your most important content pages have appropriate schema: Article for blog posts and editorial content, Product for product pages, FAQPage for FAQ content. Focus on the pages that address the questions your potential customers are most likely asking AI tools.

Then audit your entity consistency. Your Organization schema should match the information on your Google Business Profile, your social media profiles, your industry directory listings, and your Wikipedia page (if you have one). Inconsistencies between these sources weaken the signal rather than strengthening it.

Finally, don't stop at your own site. Structured data on your own domain is only part of the picture. Your presence in third-party sources that maintain their own structured data, review platforms, industry directories, comparison sites, matters just as much for how AI systems build their understanding of your entity.

Where this is headed

The relationship between structured data and AI visibility is likely to become more direct over time, not less. As AI search tools mature and compete for accuracy, they'll increasingly seek out the most reliable signals for understanding entities and verifying claims. Structured data, by its nature, provides exactly that kind of signal.

Google's continued investment in knowledge graph infrastructure, Microsoft's explicit acknowledgment of schema's role in LLM understanding, and the broader industry shift toward entity-based search all point in the same direction: brands that invest in comprehensive, accurate structured data today are building a foundation that will compound in value as AI search grows.

The debate about whether structured data "works" for AI visibility is framed too narrowly. The better question is whether you want AI systems to understand your brand accurately and confidently. If the answer is yes, structured data is one of the most direct tools you have for making that happen.

James Calder is the editor of The Search Signal, covering AI-powered search, generative engine optimization, and the future of brand discovery.