Published March 19, 2026
Google is how people find businesses on the web. Schema is how AI finds and understands businesses. Without it, you are invisible to the system.
When someone asks an AI assistant to recommend a business, the AI needs answers: what does this business do, where does it operate, can the user reach them, and are they open right now? Schema is how a website provides those answers in a format AI can read without guessing.
But not every schema field carries equal weight. Some directly influence whether AI cites a business. Others exist in the schema.org specification but do not change how AI treats a page. Understanding which fields matter is the foundation for schema that drives AI visibility.
Instead of indexing pages and ranking them against keywords, AI reads a page, builds an understanding of the business behind it, and decides whether to recommend it. That decision depends on whether the AI has enough structured information to trust what it knows.
LLM visibility — whether an AI can find and understand your business — comes before discoverability. Schema gives AI machine-readable facts so the model does not have to infer them from page content. A business without schema is invisible to AI the same way a business without a Google listing is invisible to search.
The schema fields AI systems consistently used fell into three categories.
Contact data is both a trust signal and functional data. AI treats a reachable phone number and physical address as confirmation the business is real. It also needs contact information to provide to users who ask. If it is not in schema, AI moves on to a business whose contact data is structured.
If a user asks for a recommendation at 3 PM, AI checks whether the business is open. Without hours in schema or up-to-date directories, AI cannot confirm availability and will recommend a business whose hours are verifiable.
Pricing lets AI pre-qualify users before referring them. A controlled experiment showed AI-referred traffic converting at 24.9% — roughly 6–10x the industry benchmark. Transparent pricing contributes to that qualification effect: fewer sessions, but users who arrive already know the price range.
These three categories share a common trait: they give AI what it needs to complete a transaction loop on behalf of the user. Schema is not about describing your business to a search engine. It is about giving AI the data to confidently connect a user to your business.
Across the sites in this research, 90% of issues were missing fields — not miscategorized businesses. The data was simply not there.
Credentials and authority signals were not required for initial AI citation. A separate experiment confirmed this: a brand-new site on a fresh domain — no backlinks, no social media, no reviews — earned its first AI-sourced lead in 33 days with complete schema and content that matched its schema context. Authority layers on top to increase recommendation frequency, but without foundational schema fields, authority alone does not get you cited.
Correct schema makes you eligible to appear in AI responses. It does not guarantee you will. If schema is missing, the business is invisible regardless of reputation. If schema is present, the business enters the consideration set — and from there, field accuracy, content alignment, and third-party signals like Google Business Profile determine how far it goes. Schema is not a ranking factor. It is a qualifying factor.
Schema and page content must tell the same story. If schema identifies a business as a commercial roofing contractor but the page reads like a general contractor's site, AI confidence drops. AI cross-references structured data against page content — when they conflict, it looks for a clearer source.
The relationship between page structure and AI visibility is foundational — AI reads structure first, content second. The AIFDS blueprint library is built around this principle.
Start with the homepage. The homepage is where AI forms its first understanding of the business. Sites without homepage schema were consistently unlikely to be cited.
Focus on fields that complete the loop. Contact information, hours, and service details are the data AI needs to confidently send a user to a business.
Match schema to page content. When they align, AI confidence increases. When they conflict, the page gets skipped.
Consider pricing transparency. If the business model benefits from pre-qualified leads over volume, pricing in schema lets AI filter on your behalf.
Every blueprint in the AIFDS library is built from this research — 490+ blueprints across 7 industry families and 32 business types. The fields included are the ones AI demonstrated it uses. The fields excluded are the ones that did not change AI behavior. Each blueprint is ready to deploy, paired with an AI prompt that generates the complete markup from your business details.
David Valencia writes about how AI systems find, parse, and cite websites.