Schema Markup and Entity SEO for AI Search: Help AI Understand Who You Are
Sergey Carp from New York City, Co-founder / CTO
Ask an AI assistant about your own company and read the answer carefully. Quite often it lists an address you moved out of three years ago, a price from last year's rate card or a service you stopped offering long ago. Sometimes it is worse: the AI mixes you up with a similarly named business in another city and attributes their reviews to you. People on your website understand everything from context. A machine assembles the picture from fragments it found in different places, and somewhere along the way it gets things wrong.
You can fix this by telling AI, unambiguously, who you are, what you sell, where you operate and who stands behind the business. That is what schema markup, a well-built FAQ and what SEO specialists call your company's "entity" are for. We will walk through all of it step by step, in plain English and without code.
Key takeaway
AI recommends companies it can clearly identify. Consistent facts, schema markup, a real FAQ and linked profiles make your business a clear entity.
How AI "recognizes" a company
When someone asks ChatGPT, Gemini or Perplexity "which dental clinic in New York City offers all-inclusive dental implants", they get an AI answer: a coherent text that names a few clinics, mentions prices and terms, and links to its sources. In Google, a similar answer appears as an AI Overview above the regular results; in chat assistants, it shows up right in the conversation.
To name your clinic, the AI has to be confident about three things: the organization exists, it does what the person asked about, and the information about it is reliable. That confidence comes from your website, map listings, directories, reviews and media coverage. When the sources agree, the picture is clear. When they contradict each other, the AI either makes mistakes or picks a business it understands better.
What that confusion costs a business:
- Wrong facts in answers. A customer calls an old number or shows up at an old address.
- Lost recommendations. The AI is not sure you offer implants and names a competitor whose website says so plainly.
- Someone else's reputation. Reviews of a business with a similar name get attributed to you.
Entities and the knowledge graph
Search engines and AI assistants think not only in words but in things. A thing, or entity, is a specific real-world object: a company, a person, a city, a product. An entity has properties (name, address, founding year, CEO) and relationships with other entities (the company is located in a city, provides a service, employs an expert, was covered in an article).
A network of such entities and relationships is called a knowledge graph. Major search engines maintain their own knowledge graphs, and language models also rely heavily on relationships between entities that they extract from text. That is why knowledge graph SEO and AI search optimization are directly connected: the more clearly your company is described as a distinct entity, the lower the chance that AI confuses it with someone else or leaves it out.
Entity SEO is the work of making machines recognize your company as a distinct entity with verified properties. In practice, it comes down to three things:
- consistent facts about the company everywhere it is mentioned;
- a machine-readable description of those facts on your website, in other words, markup;
- connections to other well-known entities: profiles on major platforms, publications, experts, partners.
Schema markup for AI search: what it is and whether you need it
Schema.org is a shared vocabulary for describing things on web pages, supported by Google, Bing, Yandex and other systems. Schema markup is a set of labels in your page code that visitors do not see: "this is an organization", "this is its address", "this is a service", "this is the price of the service", "this is a question and this is its answer". It is most often added in the JSON-LD format, as a separate data block in the page code.
Do you need schema markup for AI search? The honest answer: yes, but not as a magic switch. Major AI providers do not disclose how much weight they give markup when generating answers. Still, there are solid reasons to implement it:
- search engines use markup, and AI answers rely on search: Google AI Overviews and AI Mode on the Google index, Copilot on Bing;
- markup removes ambiguity: a machine does not have to guess that "from 350" is a price and not a product code;
- markup connects your website to your company's profiles on other platforms and helps assemble the entity;
- it is relatively inexpensive to implement and does not get in the way of visitors.
The main limitation: markup does not replace content. If a page has no clear description of a service, labeling it "this is a service" achieves nothing. And the data in the markup must match what people see on the page. Mismatches are an error, and search engines may stop trusting your markup because of them.
Which schema types a company needs
The Schema.org vocabulary is huge, but most companies need seven or eight types. We deliberately do not include code samples: a developer or your CMS will generate the code anyway. What matters is knowing which fields to fill in and where the data comes from.
| Type | Where to use it | Key fields | What it gives AI |
|---|---|---|---|
| Organization (or LocalBusiness) | Home, About and Contact pages | name, legalName, url, logo, address, telephone, foundingDate, sameAs | An unambiguous description of the company as an entity, linked to its profiles |
| Service | Service pages | name, description, provider, areaServed, offers with price | A clear picture of what you do, where and for how much |
| Product | Product pages | name, description, brand, sku, offers (price, priceCurrency, availability) | Accurate prices, availability and product details |
| FAQPage | Pages with a questions and answers block | mainEntity with a list of Question items and acceptedAnswer | Ready-made question and answer pairs that are easy to quote |
| Review and AggregateRating | Product and service pages with reviews | author, reviewRating, reviewBody, ratingValue, reviewCount | Quality signals tied to a specific product or service |
| Person | Expert and author pages | name, jobTitle, worksFor, knowsAbout, sameAs | A link between experts and the company, plus their authority |
| Article | Blog posts and articles | headline, author, datePublished, dateModified, publisher | Who wrote it, when it was updated and on whose behalf |
| BreadcrumbList | All inner pages | itemListElement with section names and URLs | The site structure and where the page sits in it |
Organization: your company's passport
This is the foundation of your entity. The name field holds the name customers know you by; legalName holds the registered legal name. The address field contains the full address broken into parts: street, city, postal code, country. Telephone goes in international format, and foundingDate holds the year you were founded. The most underrated field is sameAs: a list of links to the company's official profiles on maps, directories, social networks and industry platforms. It tells machines "all of these are the same organization". If you have physical locations, use the more specific LocalBusiness subtype or an industry type such as MedicalClinic or Store.
Service and Product: what you sell
For a service, what matters is a clear name, a description, a link to the providing organization in the provider field and the service area in areaServed. The price goes in offers. If your price is "starting from", you can express it as a range or a minimum value, but that price must be visible on the page. For a product, price, currency and availability are essential: these are the details AI assistants most often repeat in shopping answers.
Review: reviews without self-promotion
Mark up reviews where they relate to a specific product or service and are published on that page. Marking up glowing reviews of your company as a whole makes little sense: search engines stopped showing star ratings for self-serving reviews about an organization published on its own website a long time ago. AI assistants trust reviews on independent platforms far more, so that is where review work should happen.
Person and Article: the experts behind the content
Every article author gets a page with Person markup: job title, a link to the organization in worksFor, areas of expertise in knowsAbout, and external profiles and publications in sameAs. Articles are marked up as Article with the author and date modified. This way the "company" entity is surrounded by "expert" entities, and AI sees real people behind the brand.
FAQ schema and FAQ content: the most underrated format
People ask AI assistants questions; they do not type keywords. "How much do all-inclusive dental implants cost?", "Does getting an implant hurt?", "How long does an implant last?" If your page already asks that question and answers it clearly, that is the easiest place for AI to take the answer from. That is why FAQ content is one of the most useful formats for AI search.
A separate note on FAQPage markup. Google significantly limited FAQ rich results in regular search, and many concluded the markup is no longer needed. But for a machine, it still clearly labels where the question is and where the answer is, so it is worth keeping when the page has a genuine questions and answers block.
How to build an FAQ that works:
- Use real questions. Take them from prompt research (the list of questions customers ask AI), from sales team emails and from calls. Invented questions like "why are we the best" are useless.
- Answer in the first sentence. "Yes", "no", "from $3,500", "3–5 business days": up front, details after.
- Keep the FAQ next to the topic. Implant questions belong on the implants page, not in a general section with a hundred questions.
- Put answers in the HTML. Expandable blocks are fine if the text is already in the page code and not loaded by a script on click.
- Mark up exactly what is visible. Questions and answers in FAQPage match the text on the page word for word.
- Keep it current. An outdated answer with last year's price is worse than no answer at all.
A single source of truth for company facts
Markup and FAQs only work if their facts match the facts everywhere else. That is why we start entity work not with code but with a document: a single company fact sheet. It is one spreadsheet that holds the reference wording for everything.
What goes into the fact sheet:
- the customer-facing name and the legal name, plus company registration or tax identifiers where relevant;
- a one-line and a one-paragraph company description;
- founding year, headquarters and service area;
- addresses, phone numbers and business hours, all in one format;
- services and product categories with current prices or price ranges;
- licenses, certifications and association memberships;
- key people: names, job titles, areas of expertise;
- links to all official company profiles;
- the date each fact was last verified.
From there, it is simple. Your website, markup, llms.txt, map listings, directory profiles, social media and copy for external publications are all checked against the fact sheet. When a price or address changes, you update the fact sheet first and then every place on the list. It sounds bureaucratic, but this is exactly how the contradictions that make AI confuse your company disappear. How to confirm these facts on third-party websites is covered in our article on brand mentions and citations for AI search.
Common mistakes
- Markup from a theme. The website theme added Organization markup with a demo company name, and it has sat in the code for years.
- Mismatch with page content. The markup says $300, the page says $350; the markup shows a 5.0 rating, but there are no reviews on the page.
- Inconsistent names. "Clinic N Dental" on the website, "ClinicN" on maps and "Clinic N Dental LLC" in a directory.
- Empty sameAs. The company's profiles exist, but the website is not connected to them.
- Experts without pages. Articles are signed "Admin" or not signed at all.
- Markup instead of content. FAQPage is in place, but the answers are one word long.
How to test your markup
After implementation, run your key pages through the Schema Markup Validator and Google's Rich Results Test, and keep an eye on structured data reports in Google Search Console. There should be no errors, and warnings are worth reviewing. Then spot-check the markup data against the visible text. Once a quarter, compare the markup with the company fact sheet, especially prices and contact details.
Markup and consistent facts are a dedicated section of our AI search optimization checklist. The checklist also shows what has to come before them: crawler access and HTML that can be read without JavaScript.
The result: AI understands who you are
Once the fact sheet is in place, the website is marked up, the FAQ answers real questions and the company's profiles are linked together, your company stops being a set of scattered mentions to a machine. It becomes a clear entity with well-defined properties: name, address, services, prices, experts. There are fewer errors in answers, and AI has fewer reasons to pick a business it understands better instead of you. This does not guarantee a spot in AI answers, but it removes one of the main reasons you are not there.
How we make your company clear to AI
AI Search Optimization has been doing SEO since 2002. We have 100+ clients around the world, Google and Yandex certifications, and one team that handles your website and your AI visibility end to end. We work with companies from New York City and in international markets.
Schema markup, FAQs and the company fact sheet are part of our broader work, which includes:
- an AI visibility audit: what AI assistants say about you and where they get your facts wrong;
- prompt research: the customer questions that your FAQ grows from;
- technical fixes on the website: schema markup, llms.txt, AI crawler access, speed, error fixes;
- new content built for AI answers: service pages, FAQs, expert articles and author pages;
- brand mentions on third-party sites with consistent facts;
- publishing articles in external sources to build the authority of your company and its experts;
- ongoing AI visibility tracking.
GEO is built on top of good old SEO, and AI evaluates your website as a whole. That is why we take care of the entire site from day one, from technical fixes to GEO. We guarantee delivery of GEO tasks, error fixes and fast adaptation of your website to changing AI search requirements.
The monthly fee is $290 per month, including the first 10 hours of work. Every additional hour is $45. GEO work is billed on a time and materials basis: we invoice only the work completed and delivered that month. No discounts or free months.
Book a call. We will look at what AI assistants know about your company, find where they get it wrong and start with your company fact sheet.