AIReadyCheckMost schema markup advice is written for Google. Add Product schema for rich snippets. Add FAQ schema for those dropdown results. Add Article schema for Google News. It's all about traditional search features.
But AI models don't render rich snippets. They don't show dropdown FAQs. They read your structured data as raw text, parse the relationships between entities, and use that information to decide whether to recommend you. The schema types that earn Google rich results and the ones that actually influence AI recommendations are not the same list.
We've audited over a hundred sites for AI visibility, and we consistently see the same pattern: sites with the right structured data get cited more often by ChatGPT, Gemini, and Perplexity. But "the right structured data" is a shorter list than most agencies will tell you.
Here's what actually matters.
How AI Models Read Structured Data
Before we get into specific schema types, you need to understand how AI models interact with your JSON-LD. It's different from how Google uses it.
Google parses your structured data to trigger specific search features. Product schema triggers product panels. FAQ schema triggers accordion dropdowns. The relationship is mechanical: add this markup, get that feature.
AI models like ChatGPT and Perplexity treat structured data differently. Their crawlers (GPTBot, PerplexityBot) read JSON-LD as additional text on the page. They don't parse it into a separate database of structured facts the way Google does. Instead, the structured data becomes part of the context that informs the model's understanding of your page.
This means two things. First, your schema reinforces what's already on the page. If your visible content says you're an accounting platform and your Organization schema confirms it, that's a stronger signal than either one alone. Second, schema that exists purely for search features (like BreadcrumbList or SiteNavigationElement) adds almost nothing for AI.
One critical technical detail: most AI crawlers don't execute JavaScript. If your JSON-LD is injected client-side by a React component or a tag manager, many AI crawlers won't see it. Your structured data needs to be in the raw HTML that the server sends. Test this with a simple curl command:
curl -s https://yoursite.com | grep "application/ld+json"If nothing comes back, your schema is invisible to AI crawlers.
Organization Schema: The One Everyone Needs
This is the single most important schema type for AI visibility. It's also the one most sites either skip or implement poorly.
Organization schema tells AI models who you are, what you do, and where else you exist on the web. When ChatGPT decides which companies to recommend, it cross-references information from multiple sources. Organization schema with a solid sameAs array connects the dots.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Acme Cloud Solutions",
"url": "https://acmecloud.com",
"description": "Cloud infrastructure monitoring for mid-market SaaS companies.",
"foundingDate": "2019",
"numberOfEmployees": {
"@type": "QuantitativeValue",
"minValue": 50,
"maxValue": 100
},
"sameAs": [
"https://linkedin.com/company/acmecloud",
"https://twitter.com/acmecloud",
"https://github.com/acmecloud",
"https://g2.com/products/acme-cloud-solutions",
"https://www.crunchbase.com/organization/acmecloud"
],
"knowsAbout": [
"cloud monitoring",
"infrastructure observability",
"SaaS reliability"
]
}
</script>A few things to notice. The description field is specific: what you do, for whom. Not a tagline, not a mission statement. The sameAs array includes every legitimate third-party profile. And knowsAbout signals your areas of expertise, which helps AI models categorize you correctly.
The most common mistake we see is an Organization schema with just a name and URL. That's barely better than nothing. AI models already know your name and URL from crawling your page. What they need is the context that helps them compare you against competitors.
Product and Service Schema: What Gets You Into Comparisons
When someone asks ChatGPT "what are the best project management tools under $20 per month," the model needs to know your pricing, your features, and your category. Product and Service schema provide exactly that.
This is where structured data directly influences whether you appear in AI recommendation queries. Without it, the model has to guess your pricing from scattered text on your website. With it, the information is unambiguous.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Product",
"name": "Acme Monitor Pro",
"description": "Real-time cloud infrastructure monitoring with anomaly detection and automated alerts.",
"brand": {
"@type": "Brand",
"name": "Acme Cloud Solutions"
},
"category": "Cloud Monitoring Software",
"offers": {
"@type": "AggregateOffer",
"lowPrice": "29",
"highPrice": "199",
"priceCurrency": "USD",
"offerCount": "3"
},
"aggregateRating": {
"@type": "AggregateRating",
"ratingValue": "4.6",
"reviewCount": "342",
"bestRating": "5"
}
}
</script>For service businesses that don't have a "product" per se, use Service schema instead:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Service",
"name": "Cloud Migration Consulting",
"provider": {
"@type": "Organization",
"name": "Acme Cloud Solutions"
},
"description": "End-to-end cloud migration planning and execution for companies moving from on-premise to AWS or Azure.",
"serviceType": "IT Consulting",
"areaServed": {
"@type": "Place",
"name": "North America"
},
"offers": {
"@type": "Offer",
"price": "15000",
"priceCurrency": "USD",
"description": "Starting price for standard migration assessment"
}
}
</script>Include pricing if you can. We understand the hesitation, but AI models get asked about pricing constantly. If your competitors include theirs and you don't, guess who shows up in the answer. Including at least a "starting from" price in your structured data puts you in the running for those queries.
FAQ Schema: The AI Citation Machine
FAQ schema consistently outperforms every other type when it comes to AI citations. The reason is structural: the question-and-answer format maps perfectly to how people query AI models.
When someone asks Perplexity "how do I choose a cloud monitoring tool," the retrieval system looks for content that directly answers that question. FAQ schema makes the question-answer relationship explicit and machine-readable. It's the easiest format for an AI model to extract and cite.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "How much does cloud monitoring software cost?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Cloud monitoring software typically costs between $10 and $50 per host per month. Enterprise plans with advanced features like anomaly detection and custom integrations range from $100 to $300 per host. Most vendors offer free tiers for small deployments."
}
},
{
"@type": "Question",
"name": "What's the difference between cloud monitoring and observability?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Cloud monitoring tracks predefined metrics and alerts when thresholds are breached. Observability goes further by combining metrics, logs, and traces to help you understand why something went wrong, not just that it did. Most modern tools offer both capabilities."
}
}
]
}
</script>Two rules for effective FAQ schema. First, use real questions that your customers actually ask, not questions you wish they'd ask. Check your support tickets, sales calls, and "People Also Ask" results. Second, keep answers between 50 and 150 words. Long enough to be useful, short enough to be quotable.
Don't limit FAQ schema to a dedicated FAQ page. Add it to your product pages, your service pages, and your key landing pages. Every page that answers common questions should have FAQ markup. Our GEO checklist covers this in detail.
HowTo Schema: When You Have Step-by-Step Content
HowTo schema is valuable for one specific use case: when your page contains genuine step-by-step instructions. If you publish tutorials, setup guides, or implementation walkthroughs, this schema type helps AI models extract and present your steps clearly.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "HowTo",
"name": "How to Set Up Cloud Monitoring for Your AWS Environment",
"description": "A step-by-step guide to configuring cloud monitoring on AWS using Acme Monitor Pro.",
"totalTime": "PT30M",
"step": [
{
"@type": "HowToStep",
"name": "Install the monitoring agent",
"text": "Download and install the Acme agent on each EC2 instance you want to monitor. Run: curl -s https://get.acmecloud.com | bash"
},
{
"@type": "HowToStep",
"name": "Configure your dashboard",
"text": "Log into your Acme dashboard, navigate to Settings, and select which metrics to track. Start with CPU, memory, disk I/O, and network throughput."
},
{
"@type": "HowToStep",
"name": "Set alert thresholds",
"text": "Define alert rules for each metric. We recommend starting with 80% CPU for 5 minutes and 90% disk usage as baseline alerts."
}
]
}
</script>HowTo schema is not a must-have for every site. If you don't publish procedural content, skip it. But if you do publish how-to guides and you're not marking them up, you're leaving citations on the table.
Schema Types That Don't Move the Needle for AI
Now for the part most articles won't tell you. Several schema types that are valuable for traditional SEO do almost nothing for AI visibility.
BreadcrumbList. Useful for Google to understand your site structure and show breadcrumb trails in search results. AI models don't care about your URL hierarchy. They care about your content.
SiteNavigationElement. Tells Google about your navigation menu. AI crawlers ignore it entirely.
Article schema on its own. This is controversial, but hear us out. Article schema tells Google that a page is an article, who wrote it, and when. That triggers Google News features and knowledge panels. But AI models already figure this out from the page content itself. Article schema without meaningful author credentials or associated Person schema adds very little for AI. If you do use it, pair it with Person schema that includes the author's credentials and sameAs links. That's what AI models actually use to evaluate authority.
VideoObject. Important for YouTube SEO and Google video results. AI models like ChatGPT and Claude don't watch your videos. They read text. If your video has a transcript on the page, great. The transcript helps. The VideoObject schema doesn't.
Event. Unless you're an event platform, Event schema won't influence AI recommendations about your business. It helps Google show event listings, but AI models recommending products and services don't factor in your upcoming webinar.
This doesn't mean you should remove these schema types. They still serve their Google purpose. But if you're prioritizing your schema implementation specifically for AI visibility, focus your effort on the five types above first.
Implementation Priorities
Based on what we've seen across audits, here's the order that makes the most impact:
Priority 1: Organization schema on every page. Put it in your site-wide template. Make sure description, sameAs, and knowsAbout are filled out properly. This is your entity foundation.
Priority 2: Product or Service schema on relevant pages. Every product page, every service page. Include pricing, ratings, and clear descriptions. This is what gets you into comparison and recommendation answers.
Priority 3: FAQ schema on your top landing pages. Pick the 5 to 10 pages that matter most. Add 3 to 5 genuine questions per page. This is your citation engine.
Priority 4: HowTo schema on tutorial content. Only if you have step-by-step content. Don't force it.
Priority 5: Person schema for key team members. Especially important for thought leadership content and B2B services where expertise matters.
Three Mistakes That Waste Your Effort
Even sites that implement the right schema types often get the details wrong.
Mismatch between schema and visible content. If your Product schema says $29/month but your pricing page says $39/month, AI models notice the inconsistency. Both Google and AI systems check for this. Keep your structured data in sync with what's actually on the page.
Client-side injection only. We mentioned this earlier but it bears repeating. If your JSON-LD is added by JavaScript after the page loads, most AI crawlers never see it. GPTBot, ClaudeBot, and PerplexityBot generally don't execute JavaScript. Your schema needs to be in the server-rendered HTML. Check with curl, not with your browser's inspector. For more on how AI crawlers work, see our guide on robots.txt and AI crawlers.
Generic descriptions. A Product schema with "description": "Our innovative solution" tells the model nothing. Be specific. What does it do? Who is it for? How is it different? Your schema descriptions should be as clear as your sales pitch, because for AI, they are your sales pitch.
How to Test Your Structured Data for AI
Google's Rich Results Test validates that your schema is correct for Google features. But for AI visibility, you need a different test.
First, check that your schema is server-rendered:
curl -s https://yoursite.com | grep -c "application/ld+json"If the count is zero, your schema is either missing or only injected by JavaScript.
Second, extract and review your JSON-LD:
curl -s https://yoursite.com | grep -oP '(?<=<script type="application/ld\+json">).*?(?=</script>)' | python3 -m json.toolRead through it. Is the description specific? Is pricing current? Does sameAs include your actual profiles? Is the information consistent with what's visible on the page?
Third, ask AI models directly. Go to ChatGPT or Perplexity and ask a question that should surface your business. "What are the best [your category] tools?" or "How much does [your product] cost?" If you've implemented Product schema with pricing and the model still can't answer the pricing question, something is wrong with either the implementation or your overall visibility.
What To Do Next
Structured data for AI isn't about adding every schema type you can find. It's about adding the right ones, filling them out completely, and making sure AI crawlers can actually read them.
Start with Organization schema today. It takes 15 minutes and it's the foundation for everything else. Then add Product or Service schema to your money pages. Then build out FAQ schema on your top landing pages.
If you're not sure where your structured data stands right now, or whether AI models are actually reading it, that's exactly what an AI visibility audit tells you.
Run the free AI visibility scan to check your structured data and 9 other AI readiness signals in 60 seconds.