Everyone's arguing about whether schema markup still matters for AI search. It does. It's how engines parse what you are — your business name, your location, your hours, your reviews. That's the consensus take, and it's correct. It's also the wrong thing to be arguing about.
The question worth asking is whether anything outside your own website corroborates a word of it.
Schema is a claim. Corroboration is proof.
Here's the mechanical difference: when ChatGPT, Gemini, or Claude considers whether to recommend your business, they're not just reading your website. They're cross-referencing what they find there against what they already know about you from everywhere else — Google Business Profile data, review sites, local directories, news mentions, industry databases, social signals. Schema markup is you saying "this is who we are." Those other sources are the world confirming or denying it.
A perfectly marked-up schema tag that says you're a dentist in Portland means nothing if you've never been mentioned anywhere else the engine has learned to trust. If your Google Business Profile says something different, or if you're completely absent from it, the engine weighs that contradiction. If you exist in zero directories and have zero reviews outside your own website, the engine notices that too.
Schema doesn't help you win when you're starting from zero. It helps you win when you're already visible enough that the engine needs clarification on the details.
The mechanics of AI entity recognition.
AI engines build what's called an "entity graph" — a map of things and how they relate. Your business becomes a node on that graph only when enough independent sources mention it. Schema markup helps the engine understand what you're claiming about that node. But the node has to exist first.
Think of it this way: if a business exists on exactly one place online — its own website — it doesn't exist on the entity graph yet. The engine has no way to know if that website is real or invented. Add it to Google Business Profile: now there's corroboration. Add a few directory listings: more corroboration. A mention in a local news article: even more. Each independent mention strengthens the signal that this is a real business worth knowing about.
Schema markup makes those corroborating signals easier for the engine to parse and aggregate. But only if they exist.
What this means for your business.
If you've been focusing purely on getting your schema "perfect," you've been optimizing the last 10% of the problem. The engine doesn't care that your schema is technically flawless if it can't find independent evidence you exist.
Start here: check whether you're actually present in the places AI engines look. Google Business Profile — complete and consistent with your website. Industry directories and local listings that matter in your space. Review sites where your customers actually go. Mentions in local or industry publications. Social profiles that are active, not abandoned.
That corroboration is what gives schema markup something to clarify and amplify. Without it, schema is just data floating in space.
The irony: most businesses get this backwards. They obsess over schema and ignore the places where engines actually build their understanding of who they are. Fix the foundation — the corroboration layer — and schema starts doing real work.
Want to know whether AI engines have actually found you? Run a free VizyReport check to see whether ChatGPT, Gemini, Claude, and Perplexity can recommend your business. You'll find out fast whether the problem is schema, corroboration, or something else entirely.