
Business owners keep coming back to one question: Does schema markup help a website get cited by ChatGPT?
There’s no straightforward yes or no. Some marketers see schema markup for AI search as an important part of their visibility strategy, while others question whether it makes much difference.
The available research offers a more balanced picture. Schema has its uses, but expecting it to improve AI citations on its own is probably asking too much.
Let’s look at what Google and Microsoft say, what the research shows, and which schema types are worth considering.
First, What Is Schema Markup?
Think about a shop with clearly labelled shelves. You can quickly identify where to find different products without having to inspect everything yourself.
Schema markup does something similar for websites.
It’s a small piece of code, usually written in JSON-LD, that helps machines understand what’s on a webpage. It can describe a product, its price, an article’s author, or the person who reviewed it.
You’ve probably seen star ratings, cooking times, and other details displayed in Google search results. Structured data often makes these rich results possible.
Search engines have used schema for years. Its role in traditional SEO is well established. What’s less certain is how much AI search tools rely on it when selecting sources for their answers.
A recipe page is a simple example. Its schema might describe the ingredients, preparation time, and calories. Visitors see the recipe, while machines get additional context about its content.
You can also improve how information is organised, as explained in our guide to structuring information for AI search .
Does Schema Markup for AI Search Really Work?
It can help, but the evidence suggests its impact is mostly indirect. Adding schema won’t turn an ordinary article into something ChatGPT suddenly wants to cite.
Three findings help explain why.
What Google Says
Google’s position is fairly clear. Its May 2026 guidance on generative AI features explains that structured data isn’t required to appear in AI Overviews or AI Mode. There’s no special schema markup website owners need to add for these features.
Google still recommends supported structured data for traditional rich results. So, there’s no reason to remove a schema that’s already serving a useful purpose.
You can read Google’s official guidance on AI features and structured data.
What Microsoft Says
Microsoft has taken a slightly more encouraging position, indicating that schema can help its language models understand website content.
This is worth considering if visibility in Copilot matters to your business. However, it doesn’t prove that every AI tool uses structured data in the same way.
What the Research Shows
Ahrefs initially analysed six million URLs and found that pages cited by AI were nearly three times more likely to contain JSON-LD.
That sounds like a strong argument for schema. But the follow-up research produced a different picture.
In its second study, Ahrefs tracked 1,885 pages that added schema and compared them with 4,000 pages that didn’t.
The results were far less convincing.
ChatGPT citations changed by roughly 2.2%, while Google AI Mode citations changed by around 2.4%. Neither showed a meaningful improvement. Google AI Overviews citations dropped by 4.6%, although Ahrefs cautioned against treating this as proof that schema caused the decline.
So, what’s the explanation?
Websites with structured data often have other things working in their favour. They may publish better content, maintain stronger technical SEO, and have more established authority. Schema could simply be one characteristic these websites share.
There’s another limitation worth considering. The second study examined pages already visible to AI tools. It doesn’t establish whether adding schema helps completely new pages get discovered.
The takeaway is simple: a connection between schema and AI citations doesn’t automatically mean one causes the other. For now, the evidence that schema directly increases citations remains limited.
Which Schema Types Are Worth Adding?
You don’t need to add every schema type available. Start with the ones that accurately describe your business and content.
Here are the types worth considering.
Organization
Think of Organization schema as your company’s digital identity card.
It includes your business name, website, logo, and official social media profiles. The sameAs property connects your website with those profiles.
This can help search engines and AI systems distinguish your business from others with similar names.
If you’re new to structured data, it’s a sensible starting point.
Article or BlogPosting
If you publish articles regularly, Article or BlogPosting schema deserves attention.
Include the headline, author’s name, publication date, and last-updated date.
Author information is particularly useful. A genuine author profile with a relevant biography gives readers some background about who’s providing the information. It also makes authorship details easier for machines to interpret.
Dates matter too, especially for topics where information changes frequently.
Product and Offer
For ecommerce websites, Product and Offer schema can describe prices, availability, brands, and reviews where applicable.
Accuracy is essential. If your markup says a product is available but visitors see an out-of-stock message, you’re providing conflicting information.
Product schema also supports Google’s traditional rich results, making it useful even without a proven improvement in AI citations.
LocalBusiness
Businesses with physical locations should consider LocalBusiness schema.
It can describe your address, phone number, opening hours, and service area.
Make sure these details match your Google Business Profile and other business listings. Having different phone numbers or opening hours across platforms can confuse customers and create inconsistent information for search systems.
Getting the basic details right matters more than adding extra properties.
FAQ Schema 2026: Dead, Alive or Somewhere in Between?
FAQ schema has become a confusing subject, particularly since Google’s recent changes.
On May 7, 2026, Google stopped displaying FAQ rich results in Search. These were the expandable questions and answers that sometimes appeared beneath search listings.
Support was subsequently removed from the Rich Results Test in June and the Search Console API in August.
Does that mean FAQPage markup is completely useless?
Not necessarily.
FAQPage remains a valid schema.org type, and Bing has reportedly continued using it.
You’ll also find articles claiming FAQ schema is essential for getting AI citations. The problem is that strong evidence supporting these claims is difficult to find.
A practical approach is to keep useful question-and-answer sections on your website without automatically adding FAQ markup to every page.
If you have a dedicated FAQ page with genuine customer questions and helpful answers, using the markup may still make sense.
For most businesses, FAQ schema 2026 shouldn’t be a priority over more important SEO improvements. Focus on answering customer questions properly first.
Practical Steps You Can Take This Week
You don’t need a lengthy technical project to get started. A few checks can help you avoid common problems.
– Audit existing schema: Use the Schema Markup Validator to find outdated details, duplicate markup, and conflicting properties.
– Check page accuracy: Make sure your structured data matches the information visitors can actually see.
– Prioritise relevant types: Start with Organization and Article schema, then add others according to your website’s needs.
– Review raw HTML: Some AI crawlers may not execute JavaScript, potentially missing schema added through scripts. Ask your developer whether the markup is available in the server-delivered HTML.
– Keep articles updated: Add useful information to older content instead of changing dates just to make it appear fresh. You can also consider IndexNow to notify supported search engines about updates.
– Monitor AI visibility: Once a month, test relevant customer questions in ChatGPT, Gemini, and Perplexity. Record whether your business appears and which pages receive citations.
Manual checks won’t give you a complete picture. AI responses can vary, and a few searches aren’t enough to measure overall performance. Still, tracking the same questions regularly is better than relying on assumptions.
Common Schema Mistakes to Avoid
A few implementation mistakes can undermine an otherwise good structured data setup.
The first is marking up information that doesn’t appear on the page. Google’s guidelines require structured data to accurately represent visible content. Ignoring these requirements can affect eligibility for rich results.
Duplicate markup is another problem. Your SEO plugin, website theme, and page builder might all generate their own schema. If they describe the same business or article differently, you can end up with conflicting information.
Review markup needs attention too. Fake reviews and manipulated ratings can violate Google’s guidelines and damage customer trust.
Finally, remember that business information changes. Product prices, opening hours, and employee details don’t stay the same forever. If your structured data still reflects information from two years ago, it’s time to update it.
Treat schema as ongoing website maintenance, not a one-time installation.
What Matters More Than Schema?
Here’s something worth keeping in perspective: AI tools need to find, understand, and trust your content before they can reference it.
Schema can explain what’s on a page, but it can’t establish credibility by itself.
Our guide to separating AI search facts from GEO misinformation explores some of the most common misconceptions
Google continues to emphasise familiar SEO fundamentals for its AI search features. Answer important questions clearly, preferably near the beginning of an article. Include genuine experience where relevant, identify your authors, support factual claims with credible sources, and make sure your website is easy to crawl.
Mentions and references from reputable websites can also help establish your brand’s presence beyond its own pages.
This is where structured data SEO fits into the bigger picture. It supports your website’s technical foundation, but it cannot replace original research, useful explanations, or information readers can’t easily find elsewhere.
Even perfectly implemented schema won’t guarantee AI citations if your content gives search tools no particular reason to reference it.
The Bottom Line
Schema markup for AI search is worth implementing, but it’s important to understand what it can and cannot do.
Start with Organization, Article, and other schema types that genuinely describe your website. Keep the information accurate and consistent with what’s visible to visitors.
Use FAQ markup where it makes sense, particularly on dedicated FAQ pages, but don’t expect FAQ schema 2026 to transform your AI visibility.
The research is still developing, and Google and Microsoft have different positions on structured data.
For now, focus on getting the technical basics right and publishing content that’s genuinely useful and worth citing. Be cautious of anyone promising guaranteed AI citations simply because you’ve added schema.







