Published March 19, 2026
The schema.org specification includes hundreds of fields across dozens of types. Conventional wisdom in the SEO industry is to include as many as possible — the more data a page provides, the better. For traditional search engines, that advice has some merit. For AI systems, it does not hold up.
The AIFDS blueprint library includes only the fields AI systems have been shown to parse and use. Every excluded field was excluded for a reason — it did not change AI citation behavior in testing. This article explains which fields were left out and why.
The strongest evidence for which fields AI ignores comes from a controlled experiment designed to test exactly this question. A brand-new website was launched on a freshly purchased domain with a single purpose: to see whether a site with minimal signals could earn AI citations and convert real users.
The site had none of the signals the SEO industry treats as essential:
What it did have: complete schema on the homepage with the fields that matter — business type, contact information, a clear description of what the site does, and content that matched the schema context.
Within 60 days, the site was generating an average of 5 submissions per day. AI-referred traffic converted at 24.9% — roughly 6–10x the industry benchmark. The first AI-sourced lead arrived just 33 days after launch. By the second month, ChatGPT was the second-largest traffic source behind direct — ahead of Google, Bing, DuckDuckGo, and Yahoo combined.
The signals that drive traditional SEO — domain authority, backlinks, review profiles — were completely absent. The site succeeded because the fields it did include were the ones AI actually uses. Everything else was noise.
Each of the following fields is commonly recommended in schema markup guides. None of them changed AI citation behavior in the sites tracked across this research.
The experiment site had zero social media presence and earned citations within 33 days. Social links may have value for brand verification over time, but they are not required for AI to cite a business.
Reviews are a major signal for Google search results. For AI citation, the data showed they are not a deciding factor. The experiment site had no reviews of any kind and still converted at 24.9%. AI evaluates whether a business can fulfill a user's need, not whether other users have rated it highly.
When a business was founded does not help AI answer the question it needs to resolve: can this business help this user right now? The experiment site was days old. AI cited it anyway. Founding date is biographical information, not functional information.
Employee count tells AI nothing about whether the business can serve the user. A solo contractor and a 500-person firm can both fix a roof. AI needs to know what the business does, where it operates, and how to reach it — not how many people work there.
The experiment site had no credentials, no expert backing, no institutional affiliation — and earned its first AI-sourced lead in 33 days. Credentials are not worthless, but they are not required for AI citation.
The pattern across every excluded field is the same: none of them help AI complete the transaction loop. AI needs to answer: What does this business do? Where is it? Can the user reach it? Is it available right now? Those questions map to the fields that matter for AI citation — business type, contact information, hours, service area, pricing.
The excluded fields answer different questions — founding date, employee count, awards, ratings. Valid questions, but not the ones AI resolves when deciding who to recommend. AI is solving a completion gap: it needs a business that can do the thing the user is asking about, in the right place, right now.
Social links may help with brand verification as AI evolves. Reviews may influence recommendation frequency once a business is already in the citation pool. Credentials may strengthen trust for users who visit the site after AI sends them there. These fields have value — but they should not be prioritized over the fields AI needs to make a recommendation.
A business that adds awards and review schema to every page but has no contact information or hours in schema has the priority inverted. LLM visibility comes first. Get the foundational fields right — the ones that make the business findable and understandable to AI. An AI-recommendable organization starts with the basics, not the extras.
Every blueprint in the AIFDS library was built with these findings in mind. The blueprints prioritize the fields that make a business eligible for AI citation over fields that look comprehensive on paper but do not change AI behavior. The result is leaner, more focused schema — just the fields AI needs, organized by industry and page type, ready to deploy.
David Valencia writes about how AI systems find, parse, and cite websites.
The companion article covering which fields matter.
Read →