What Schema Types Do AI Crawlers Actually Read?
PerplexityBot and ChatGPT-User are real crawlers visiting your site right now. The structured data they find determines whether your business becomes an AI-cited entity or an anonymous website that gets paraphrased without attribution. Most businesses have schema. Very few have schema that AI actually acts on.
Not sure what schema your site actually has or whether AI crawlers can read it? The free Blind Spot Report includes a full schema audit.
- How AI Crawlers Differ from Googlebot
- The Schema.org Hierarchy and Entity Recognition
- The 5 High-Value Schema Types for AI Visibility
- Why JSON-LD Wins Over Microdata for AI Parsing
- Entity vs. Website: The Distinction That Determines Citations
- sameAs Links and Cross-Platform Identity Signals
- Schema Types That Don't Move the Needle
- Why Incomplete Schema Can Be Worse Than No Schema
- Schema Types Ranked by AI Citation Impact
- Frequently Asked Questions
How AI Crawlers Differ from Googlebot
Most web professionals understand Googlebot. It crawls pages, indexes content, and feeds Google Search. What fewer people understand is that AI companies run their own separate crawlers with different priorities and different data extraction behavior.
PerplexityBot and ChatGPT-User are named, documented web crawlers. They appear in your server access logs. They respect robots.txt directives. They crawl on their own schedules, independent of Google's index. And they are looking for something specific when they arrive at your site: structured data they can use to understand, categorize, and potentially cite your business in AI-generated answers.
Both PerplexityBot and ChatGPT-User are real, active crawlers documented in their respective companies' crawler policy pages. If you check your server access logs, you will likely find entries from both. They are not crawling to rank your site in a traditional search index. They are building the data layer that determines which businesses their AI systems cite when users ask questions like "find me a good HVAC company in Denver" or "who does commercial roof repair in Atlanta."
Googlebot has always prioritized visible page content: headlines, body text, internal links, and page speed. AI crawlers prioritize structured data first. When PerplexityBot hits a page with clean JSON-LD, it extracts that structured payload in milliseconds and uses it to classify the entity behind the page. When it hits a page with no schema or broken schema, it falls back to prose interpretation, which is slower, less reliable, and far less likely to produce an accurate entity classification.
The practical consequence: a business with well-implemented schema that no one would call a design-forward website can consistently outperform a beautifully built site with no structured data in AI citation rates. For more on how AI systems evaluate and decide to visit websites in the first place, see our guide on what your website looks like to an AI crawler.
Want to confirm whether PerplexityBot or ChatGPT-User are crawling your site and what they are finding? Call (213) 444-2229 for a quick crawl diagnostic.
The Schema.org Hierarchy and Why It Matters for AI Entity Recognition
Schema.org is organized as a hierarchy. At the top sits the generic type called Thing. Everything in the vocabulary is a subtype of Thing. Below Thing sits Organization, which is the parent of LocalBusiness, which is the parent of dozens of specific business categories: Restaurant, Dentist, HomeAndConstructionBusiness, ProfessionalService, Store, and many more.
This hierarchy matters for AI entity recognition because AI systems use schema type specificity to match businesses to queries. A business using the generic LocalBusiness type competes for every local query. A business using the specific Dentist type sends a direct categorical signal that focuses AI citation behavior on dental queries. A business using Plumber tells AI systems precisely which service category it belongs to without requiring the AI to interpret prose descriptions that may be imprecise or marketing-heavy.
Using @type: LocalBusiness when Schema.org has a specific type for your category is not a safe default. It is a signal that tells AI systems your business fits into a broad, undifferentiated bucket. Your competitor who uses @type: Plumber or @type: LegalService sends a precise categorical signal. In a tie for citation worthiness, the specific type wins because it matches the query more directly. Check Schema.org for your specific business category before finalizing your LocalBusiness implementation.
The hierarchy also determines which properties are valid for your schema type. Properties that are defined for a parent type are inherited by all subtypes. Properties defined for a specific subtype are only valid for that type and its children. Using a property on the wrong schema type does not break the schema, but AI crawlers that validate schema may lower confidence in a block that contains type mismatches.
Not sure which Schema.org type applies to your specific business? Email support@theanswerengine.ai and we will identify the correct type for your category.
The 5 High-Value Schema Types for Local Business AI Visibility
Schema.org has over 800 types. For local and service businesses competing for AI citations in 2026, five schema types do the meaningful work. Understanding what each one signals to AI systems, and where most implementations go wrong, is the difference between schema that gets cited and schema that decorates your source code.
1. Organization: The Entity Foundation
Organization schema is the entity foundation. It tells AI systems and knowledge graphs: this website belongs to a named organization. It establishes the entity's official name, official URL, logo, contact information, and, most critically, its sameAs links connecting it to other trusted platforms. Every business website should have Organization schema, even if more specific types are also present.
The most common Organization schema mistake is omitting sameAs entirely. A business entity that only describes itself on its own website is unverifiable. An entity that connects to its Google Business Profile, LinkedIn, Yelp listing, and relevant industry directories gives AI systems a cross-platform identity web to validate against. The sameAs property is what transforms a website into a knowledge graph node.
2. Article and BlogPosting: Content Trust Signals
Article and BlogPosting schema signal editorial authority for content pages. When PerplexityBot or ChatGPT-User encounters a page with Article schema, it receives explicit metadata: who wrote this, when it was published, when it was last updated, and what organization published it. These signals factor into the trust weighting AI systems apply when deciding whether to cite a piece of content.
The common mistake with Article schema is omitting the author entity. A floating article with no linked author is a lower-trust signal than an article where the author has their own schema entity, or where the publisher Organization schema is clearly referenced. The author field should point to a Person or Organization entity, not just a plain text name string. For a deeper look at how content trust affects AI citation behavior, see our guide on how to create content that ChatGPT actually trusts.
3. FAQPage: Direct Q&A Extraction
FAQPage schema has the most direct influence on AI citation behavior of any schema type. AI language models are designed to answer questions. FAQPage schema provides pre-structured question-and-answer pairs that retrieval systems can extract without interpretation. When ChatGPT-User finds FAQPage schema, it has the exact data format it needs to generate a sourced answer.
The questions in FAQPage schema must be phrased as actual user queries, not business talking points. "What is your cancellation policy?" is a business FAQ. "Can I cancel my HVAC service appointment same-day in Phoenix?" is an AI query. The second format matches how users ask questions and how AI retrieval systems score relevance. Minimum five questions per page, phrased as natural language queries, with substantive answers that include local specifics: city, price range, timeline, and qualifying conditions.
4. LocalBusiness with Specific Subtypes: Geographic Entity Signals
LocalBusiness schema ties the entity to a specific geographic market. It is the structured data that tells AI: this business operates at this address, in this city, serving this radius, during these hours. For any query with local intent, LocalBusiness schema is the gating signal. Without it, AI systems have no structured basis for including a business in geographically scoped recommendations.
But the subtype matters as much as the type. Use the most specific Schema.org subtype that applies: Dentist, Plumber, Electrician, LegalService, FinancialService, Restaurant, and dozens more sit beneath LocalBusiness in the hierarchy. The subtype sends a category signal that generic LocalBusiness cannot match. If your business category lacks a specific Schema.org subtype, use LocalBusiness with a detailed knowsAbout array to compensate.
5. Product: For E-Commerce and Service-Based Product Pages
Product schema extends AI visibility into transactional queries. For e-commerce sites and service businesses that sell defined packages, Product schema with an associated Offer (including price and currency) gives AI systems the exact information they need to answer "how much does X cost" queries with a sourced citation.
The most common Product schema mistake is using it without an Offer. A Product without a price is dramatically less useful to AI retrieval systems evaluating transactional queries. Even a price range expressed as a string in the priceRange property on LocalBusiness schema provides more signal than a Product entity with no cost information. If you cannot publish exact pricing, publish a price range.
Want to know which of these five schema types your site is missing? The free Blind Spot Report includes a schema type audit.
Why JSON-LD Wins Over Microdata for AI Crawler Parsing
Schema markup can be implemented in three formats: JSON-LD, Microdata, and RDFa. Google officially recommends JSON-LD, and both PerplexityBot and ChatGPT-User are optimized to parse it. The performance gap between JSON-LD and the alternatives is significant enough to matter in practice.
- Lives in a single, isolated script block in the page head
- Extractable without DOM traversal or HTML parsing
- Errors are localized: one broken block doesn't corrupt others
- Easily validated with a single JSON parse step
- Can be updated without touching visible HTML
- Google's official recommended format, AI crawlers follow Google's lead
- Supports the full @graph pattern for linked entity declarations
- Readable by any JavaScript-capable crawler without render
- Requires full DOM traversal to extract structured data
- Schema properties scattered across HTML elements throughout the page
- Template or CMS changes can silently break schema properties
- Errors are distributed and harder to detect with validators
- Slower to parse, especially on large or complex pages
- Not recommended by Google for new implementations since 2016
- AI crawlers optimized for JSON-LD may partially skip Microdata
- Mixing Microdata and JSON-LD on the same page creates parser ambiguity
The @graph pattern in JSON-LD is particularly powerful for AI citation purposes. It allows you to declare multiple related entities in a single script block and link them together using @id references. Your Organization entity can link to your LocalBusiness entity, which links to your Service entities, which link back to the Organization as the provider. This interconnected entity graph gives AI systems a complete, parseable picture of your business in a single crawl.
The most efficient JSON-LD implementation uses a single @graph array containing all your entity declarations. This tells AI crawlers: here is the complete structured data for this page, all in one place. Entities within the @graph can reference each other by @id. A crawlers that processes the @graph gets a fully linked entity picture in a single pass. Businesses with linked @graph implementations provide richer entity context than businesses with separate, unlinked schema blocks, even if both contain the same schema types.
Entity vs. Website: Why AI Treats Them Differently
The most important conceptual shift in AI search strategy is understanding the difference between a website and an entity. A website is a technical object: a domain with pages hosted at a URL. An entity is a recognized, named thing in the world: a business with a location, a history, a set of services, and a presence across multiple platforms.
AI systems and knowledge graphs track entities, not websites. When a user asks ChatGPT to recommend a plumber in Denver, the model draws from its training data and live retrieval index to identify entities that match the query. Entities that are well-established across multiple signals, including schema markup, directory listings, review platforms, and third-party mentions, are cited by name. Anonymous websites that have not established entity-level signals are paraphrased without attribution or skipped entirely.
Businesses that have established a recognized entity, not just a website, have the currency of AI-powered local search. An entity has a name, a location, a category, a set of verified signals across platforms, and a node in the knowledge graph. Once a business reaches entity status in the knowledge graph, AI systems cite it as an authoritative source. Businesses that exist only as a website without cross-platform entity signals are treated as unverified sources that AI systems cite cautiously or not at all. Schema markup is the primary mechanism for establishing entity status from your own website.
The knowledge graph is the data structure underlying Google's understanding of entities in the world. A business that appears in the knowledge graph is recognized as a real, specific entity with verified attributes. Schema markup is one of the signals that builds a business's knowledge graph node, alongside Google Business Profile data, third-party citations, Wikipedia mentions for larger brands, and consistent NAP data across directories.
For a comprehensive look at how AI systems evaluate your site's overall trustworthiness beyond schema, see our guide on how schema markup influences AI search citations.
Not sure whether your business has entity-level status or is still just a website to AI? The free Blind Spot Report checks your entity establishment signals.
sameAs Links: Why Cross-Platform Identity Signals Amplify Schema Impact
The sameAs property in Organization and LocalBusiness schema accepts an array of URLs pointing to the same entity on other trusted platforms. Your Google Business Profile listing, Yelp page, LinkedIn company page, BBB profile, and relevant industry directory listings are all candidates for sameAs entries.
The mechanism behind sameAs amplification is entity resolution. When an AI system or knowledge graph processes a sameAs array, it follows each URL and verifies whether the entity described there matches the entity described in the schema. Consistent NAP data (name, address, phone) across all sameAs targets raises confidence in the entity's authenticity. Inconsistent data lowers confidence and can prevent knowledge graph node formation entirely.
| Platform | Priority | Why It Matters for AI | Consistency Requirement |
|---|---|---|---|
| Google Business Profile | Critical | Direct Google Knowledge Graph input | Exact NAP match required |
| Yelp Business Page | High | Major AI training data source, Perplexity citation source | Exact NAP match required |
| LinkedIn Company Page | High | Professional entity signal, high domain authority | Business name and URL must match exactly |
| BBB Profile | Medium | Trust authority signal, frequently cited by AI | Accreditation status and address must be current |
| Industry Directories | Medium | Category-specific authority signals | Consistent with primary NAP |
| Chamber of Commerce | Supporting | Local authority and geographic specificity | Active listing, not expired membership |
The compounding effect of sameAs links is documented in entity resolution research: each additional trusted platform that confirms the same entity data increases the probability of knowledge graph node formation. A business with six consistent sameAs links is substantially more likely to have a knowledge graph entity than a business with two. And businesses with knowledge graph entities are cited by AI as authoritative named sources, not just referenced as anonymous web results.
Want to see which sameAs links you are missing and which ones have inconsistent NAP data? Call (213) 444-2229 for a citation entity audit.
Schema Types That Don't Move the Needle for AI Citations
Knowing which schema types to prioritize is only half the picture. Understanding which schema types are commonly implemented but rarely influence AI citation behavior helps you allocate implementation effort correctly. The following types appear frequently on local business sites and have minimal impact on AI search visibility.
| Schema Type | AI Citation Impact | Best For | Common Mistake |
|---|---|---|---|
| Organization | High | Entity foundation for all pages | Missing sameAs array, no @id linking |
| LocalBusiness | High | All local and geographic queries | Using generic type when specific subtype exists |
| Article / BlogPosting | High | Content pages, informational queries | Missing author entity, no dateModified |
| FAQPage | Medium-High | Direct answer extraction by AI | Duplicate questions, answers too brief |
| Product | Medium | Transactional queries, e-commerce | Missing Offer with price and currency |
| BreadcrumbList | Supporting | Site structure, navigation signals | Wrong hierarchy, skipping intermediate pages |
| WebPage / WebSite | Minimal | Basic page identification only | Added in place of more specific types |
| ImageObject | Minimal | Rich results for image search only | Used as primary schema without entity types |
| SiteLinksSearchBox | None for AI | Google SERP feature only | Prioritized over entity-building schema |
| VideoObject | Minimal | Video search results only | Used without video transcript schema |
The pattern is consistent: schema types that describe real-world entities and their relationships produce AI citation impact. Schema types that describe technical web objects (websites, images, videos) without entity context produce minimal impact because AI systems are trying to understand what business is behind a site, not what technical format the site uses.
Why Incomplete Schema Can Be Worse Than No Schema for Some AI Systems
There is a common assumption that any schema is better than no schema. For most situations, that is true: incomplete but accurate schema gives AI crawlers some entity data to work with, even if the picture is partial. But in two specific situations, incomplete or inaccurate schema creates problems that no schema avoids.
AI crawlers extract schema data literally. If your LocalBusiness schema contains a phone number that differs from the number on your Google Business Profile, that conflict is a red flag in entity resolution. If your schema claims your hours are 9 to 5 Monday through Friday but your GBP says you are open on weekends, the conflict reduces AI confidence in both sources. Schema that accurately describes your business is a citation accelerator. Schema that contains inaccurate or outdated data is an entity integrity problem that can suppress citations more effectively than having no schema at all.
The second situation where incomplete schema causes harm is duplicate or conflicting schema on the same site. Multiple LocalBusiness blocks with different phone numbers, multiple Organization entities with different sameAs arrays, or FAQPage schema containing duplicate questions all create parser ambiguity. Some AI crawlers resolve the conflict by lowering trust in the entire schema payload. Others pick one version arbitrarily, which may not be the accurate one.
Valid JSON-LD with accurate data and no conflicts is the baseline. Schema validation tools can catch syntax errors. But only a manual audit comparing schema data to your actual business information, Google Business Profile, and directory listings can catch the accuracy and consistency issues that suppress AI citations without showing up in a validator.
For a broader view of how your entire website structure affects AI crawlability, see our guide on what your website looks like to an AI crawler.
Concerned your schema may have accuracy or conflict issues? Email support@theanswerengine.ai for a schema integrity check.
Find Out Which Schema Types Are Missing from Your Site
The free Blind Spot Report audits your schema markup: which types are present, which required properties are missing, whether your JSON-LD is valid and accurate, and how your entity signals compare to businesses currently appearing in AI responses for your category and market.
Get Your Free Blind Spot ReportFrequently Asked Questions
Do PerplexityBot and ChatGPT-User actually crawl my website for schema?
Yes. Both are real, named web crawlers documented in their companies' crawler policy pages. They appear in server access logs, respect robots.txt, and prioritize JSON-LD structured data at every crawl. If your robots.txt blocks these crawlers or your schema is invalid, they fall back to unstructured content extraction, which is slower and less reliable for citation purposes. Check your crawl accessibility free.
Which schema types matter most for AI citations in 2026?
The five schema types with the highest documented impact are: Organization (entity foundation with sameAs links), Article or BlogPosting (content authority signals), FAQPage (direct Q&A extraction by AI retrieval), LocalBusiness with specific subtypes (geographic entity signals), and Product with Offer (transactional query matching). FAQPage has the most direct influence because AI systems are designed to extract structured question-and-answer data. Call (213) 444-2229 for a schema type audit.
Why does JSON-LD outperform Microdata for AI crawlers?
JSON-LD lives in a separate script block and does not require DOM traversal. AI crawlers extract the entire structured data payload in a single parse. Microdata requires crawlers to walk every HTML element, which is slower and more sensitive to template changes. Google officially recommends JSON-LD, and both PerplexityBot and ChatGPT-User are optimized for it. Broken Microdata is also harder to detect because errors are distributed across the HTML. Email us to audit your current schema format.
What is the difference between an entity and a website in AI search?
A website is a technical object hosted at a domain. An entity is a recognized named thing in the world: a business with a location, a set of services, and consistent presence across platforms. AI systems track entities, not websites. Entities get cited by name. Anonymous websites get paraphrased without attribution or ignored. Schema markup is the primary mechanism for establishing entity status from your own website. Check your entity establishment signals free.
What are sameAs links and why do they amplify schema impact?
The sameAs property in Organization and LocalBusiness schema accepts URLs pointing to the same entity on other trusted platforms: Google Business Profile, Yelp, LinkedIn, BBB, and industry directories. When AI systems and knowledge graphs find the same entity described consistently across multiple trusted sources, they raise confidence in the entity's legitimacy and assign it a stronger knowledge graph node. A business with five consistent sameAs links is substantially more likely to be cited by AI than one with no cross-platform signals. Call (213) 444-2229 to audit your sameAs coverage.
Can incomplete schema hurt my AI search visibility more than no schema?
Yes, in specific situations. Inaccurate schema with data that conflicts with your Google Business Profile or directory listings creates entity resolution conflicts that lower AI confidence across all sources. Duplicate schema blocks with conflicting data create parser ambiguity. Invalid JSON-LD causes crawlers to skip the block entirely, which is neutral. Inaccurate schema is actively harmful. Maintain accuracy and consistency as the highest priority, before adding more schema types. Email support@theanswerengine.ai for a schema integrity check.
More questions about schema and AI citations? Call (213) 444-2229 or email support@theanswerengine.ai.
Schema Is the Machine-Readable Layer That Determines Whether AI Cites You
The free Blind Spot Report checks your schema validity, identifies missing types and properties, audits your sameAs coverage, and compares your entity signals to businesses currently appearing in AI answers for your category and market. Takes 60 seconds to request. No credit card required.
Get My Free Blind Spot ReportNext move // Free audit
See where answer engines miss your business
Get a clear view of how ChatGPT, Claude, Perplexity, and Google answer questions about your category today.
Run the free audit