Schema markup is structured data I add to a website so search engines and AI assistants understand what each page contains. It uses a standard format that tells them whether I am describing a service, a location, a professional, an article, or something else. But schema alone does not guarantee that AI assistants will find your entire business. They also need your site to have clean structure and an explicit map of what you offer.
A few months ago I watched an AI assistant answer a question about baby sleep by finding one article on a parenting site, summarizing it well, and never mentioning that the site had dozens of other resources on the same topic. The site had the right data in the right format. The AI still found only one page. I realized the site looked good to a search engine but not to an AI looking for the whole picture.
This happens to service businesses constantly. When an AI gets asked about a dental practice, I have seen it pull up their homepage or their cosmetic services page. It has no idea they also do implants, root canals, or pediatric work. A real estate team's brokerage page gets quoted, but their individual agents never get recommended. An AI finds your location in Los Angeles and never knows you have an office in Seattle. I notice this every time I audit a site for AI discovery.
I thought schema markup would solve this. It did not.
What schema markup does
I use schema markup to tag information on a website so that machines read it the same way every time. I mark up a location with address, phone number, and hours. I mark up a person or professional with credentials and bio. I mark up a service with description, duration, and price range. These tags live in the code of your page, invisible to a visitor, but readable to a search engine or an AI assistant.
When I do this right, an AI does not have to guess what your page is about. It knows. This beats the alternative, which is that AI crawls your site, reads the text like a human would, and makes its best inference. That usually works, but inferences can be wrong. Schema markup is explicit.
The problem with relying only on markup
But here is what I discovered while building a content site that publishes hundreds of articles automatically: having good schema markup on every page is necessary, but it does not guarantee that an AI will find your whole site. When the site has no clear structure, and the AI lands on one article, it has no way to know what else exists.
The site I built, BabyJunctions, publishes articles on baby sleep, development by age, how-tos for parents, interactive tools, and buying guides. I put proper markup on each article. But when an AI assistant is asked about sleep training and lands on one sleep article, it may never crawl through to the articles on development stages, the tools, or the guides. I realized the structure of the site itself has to invite that navigation. The AI has to understand what else is there.
How I made the full business discoverable
I solved this in two ways, both practical.
The first was making sure every page on the site has the same structure and the same kind of organization. When I build each page the same way, an AI assistant can navigate it predictably. It can find a link to my services page from my homepage. It can see that services are organized by type. It can follow the pattern. Inconsistent site structure, even with perfect markup on each page, leaves AI guessing.
The second was explicit. I created a map: a file that lists every section of the site, what it covers, and where to find it. I publish this map for AI assistants to read. It costs me an hour to create and a little time to maintain, but it means an AI does not have to crawl and infer. It can read the map and know exactly what is available.
For BabyJunctions, I named this file llms.txt. When an AI assistant reads it, it learns that the site has buying guides for specific age ranges, how-to articles, learning stage guides, and about twenty free tools. The file points to each section. Instead of an AI guessing from one article, it can see the whole landscape.
What to check on your own site
If you run a medical practice, a dental office, a real estate team, or a home services business, I would ask whether an AI assistant could navigate your site the same way you can.
- Do all your service pages have the same layout and the same kind of information? Or does your cosmetic dentistry page look different from your general dentistry page?
- If an AI lands on your homepage, can it find a clear path to your list of services, locations, or team members?
- If an AI finds one team member's page, can it discover the others?
- Is the hierarchy of your site obvious from the code and the links, or would an AI have to infer it?
I use schema markup to make each page clear. But clear pages are not the same as a clear site. You need both.
The practical step
I start with schema markup if a site does not have it. That is the foundation. But I do not stop there. I audit the site's structure. I walk through it as if I were an AI that only understands links and headlines and the order of information. Does the path from one page to another make sense? Can I find everything, or do some pages feel isolated?
Then I consider whether the site needs an explicit map. A small practice with five service types and one location may not need one. A practice with three locations, fifteen service types, multiple providers, and years of patient education articles might. The question I ask is whether an AI crawling from a homepage link will naturally discover everything important, or whether some offerings will stay hidden.
I learned this the hard way by watching where AI assistants landed and what they missed. An AI can read perfect markup on a single page and do nothing with the rest of your business. What matters is whether it ever gets to the other pages. That is a question about your site's shape, not its tags.
