Schema Markup Did Not Improve AI Citations: A Practical Guide
Schema Markup for AI Optimization sounds like a straightforward route to greater visibility in AI search. Yet an Ahrefs study found that adding JSON-LD schema did not meaningfully increase AI Citations from Google AI Overviews, Google AI Mode, or ChatGPT in real-world use.
A Practical Guide to Schema Markup and AI Citations
This result does not make schema useless. Instead, it demonstrates that structured data alone can not persuade an AI system to cite a page. I need to distinguish correlation from causation while examining the written material, authority, technical SEO, and user signals behind each result.
In this article, I explain what the study found and why I do not treat Schema Markup for AI Optimization as a shortcut in practice. Instead, I focus on stronger factors that may strengthen trustworthy content and improve AI Citations over time.
Why Schema Markup For AI Optimization Did Not Improve AI Citations: A Practical Guide
Schema markup gives search engines structured specifics about a page, including products, reviews, articles, and organizations. Even so, JSON-LD alone did not produce a major increase in citations from Google AI Overviews, Google AI Mode, or ChatGPT.
Schema Markup and AI Citations
This distinction shows why correlation does not establish causation. Pages with schema might receive greater citations because they contain stronger content, better technical SEO, quality links, greater authority, and regular maintenance. Schema can help to appear on a cited site without causing that citation.
I am Anatoly Zadorozhnyy, an SEO and digital marketing expert who has assisted businesses grow through organic search since 2008. My work has shown me how search has evolved into AI-powered discovery in real-world use. To Improve AI Citations, I prioritize helpful information and trusted signals over complex updates without measurable impact.
Schema remains helpful for clarity and search presentation. It should support a sound SEO foundation, not replace solid writing, expert knowledge, or a well-maintained website. These elements give AI systems reliable material to understand and reference.
What The Ahrefs Study Found About Schema Markup And AI Citations
I reviewed the Ahrefs study by comparing pages that added Schema Markup with similar pages that did not in many cases. The research examined established pages that already had meaningful visibility in AI results. This design showed citation shifts additional clearly than a simple before-and-after count.
Main Study Sample And Comparison Method
Ahrefs tracked 1,885 web pages that added JSON-LD Schema Markup between August 2025 and March 2026 in many cases. Each page was matched with approximately three control pages from different domains. That control group contained roughly 4,000 pages with similar citation levels before the study began in real-world use.
The analysis covered Google AI Overviews, Google AI Mode, and ChatGPT. It included pages with effective prior visibility, including pages that received more than 100 Google AI Overview citations in February 2025.
Main Reported Citation Changes By Platform
Google AI Overviews recorded a 4.6% decline among pages that added Schema Markup, compared with matched control pages in practice. Google AI Mode showed a 2.4% increase, but the change was statistically indistinguishable from zero in practice.
ChatGPT showed a 2.2% increase, which was statistically indistinguishable from zero in real-world use. Such figures describe citation movement across three platforms during the study period. They do not measure every form of search performance or the full value of structured data.
Why The 4.6% AI Overview Decline Does Not Prove Schema Was Harmful: A Practical Guide
I would not treat the 4.6% decline in AI Overview citations as proof that schema markup caused harm in many cases. Pages that received schema, along with matched control pages, were already losing citations before the change in real-world use. This initial gap makes the result harder to interpret, including its implications for AI Optimization.
Treated pages declined slightly faster than control pages in practice. The average difference was about 12 fewer daily citations per site. Many sampled pages received hundreds of citations, so this gap requires context when assessing AI Optimization performance.
Ahrefs indicated that the relative decline was statistically significant. Its estimate suggested that a gap this size might occur by chance about once in 2,500 cases. Statistical significance measures the strength of a pattern, but it does not identify the cause in real-world use.
Several factors was able to explain the change. Google can have adjusted its AI Overview systems, or the content may have become stale. Page quality, major Google updates, and delayed recrawling may have affected the results. These variables may shape AI Optimization outcomes without showing that schema was harmful.
For that reason, I view the 4.6% decline as an observed difference, not a direct cause-and-effect finding in practice. The data shows that treated pages fell slightly faster, but it cannot establish why.
How Ahrefs Isolated The Effect Of Adding Schema: A Practical Guide
I treat AI Schema as a variable that requires controlled testing in many cases. A page can gain citations because a platform changes, rather than because the page received JSON-LD. Ahrefs used matched pages to distinguish schema effects from broader shifts in AI search in practice.
Matched Difference In Differences Analysis: A Practical Guide
The main strategy compared pages that received schema with similar pages that did not. Each treated webpage had a control page with related characteristics and a comparable citation pattern.
Ahrefs measured citation adjustments during the 30 days before and after the schema treatment date. This analysis accounted for platform-wide movement across AI Mode and AI Overviews.
That approach avoided a weak before-and-after comparison. A simple comparison could credit AI Schema for a trend affecting many pages simultaneously.
Four Tests That Pointed To The Same Result Explained
Ahrefs used four checks to test the stability of its analysis in practice. Each method examined citation movement from a different angle:
- One average citation change comparison used a two-sample t test.
- One difference in differences model compared treated and control pages over time.
- A week-by-week event study tracked changes around the treatment date.
- A symmetrical before-and-after test excluded the recrawling period.
Using several tests reduced the risk of relying on one model in practice. It gave the study a structured way to determine whether AI Schema produced a measurable citation lift after controlling for platform trends.
Why AI Cited Pages Are More Likely To Have Schema Markup Explained
Ahrefs indicated that around 53 percent of pages cited by AI systems used JSON-LD. Cited pages were around three times greater likely to contain schema markup than uncited pages. This pattern helps explain why Schema SEO appears in discussions of AI visibility.
That association does not prove that schema caused the citations. Organizations using structured data often pursue a broader search strategy in many cases. They might publish clearer content, maintain pages, invest in technical SEO, and earn quality backlinks.
These sites can have stronger brands and higher rankings in traditional search. Their content can earn trust from users and search systems. Together, these signals might assist an AI system retrieve and assess a page.
I view Schema SEO as one trait of a well-managed website, rather than a stand-alone method for increasing AI citations. A webpage can help to contain valid JSON-LD yet lack useful answers, original information, or clear evidence.
The key distinction is between association and cause. Schema might appear more often on cited pages because those pages belong to sites with stronger overall optimization. This distinction places Schema SEO within a broader strategy for content and authority.
What Schema Markup Still Does For SEO And AI Optimization
Schema markup retains a useful role in search. I use it to clarify site meaning, rather than present it as a direct route to more AI citations. When structured data matches visible page copy, it may sharpen search engines’ interpretation of a page.
Benefits Beyond AI Citations
Accurate schema can help to support Google rich results when a page meets eligibility requirements. It might refine displays for articles, products, reviews, local businesses, and organization details. These formats can make search results easier to scan and understand.
Schema can describe entities with greater precision. Product attributes, business details, and article information provide search systems with useful context. This clarity can assist knowledge graphs, voice assistants, and downstream entity recognition.
These gains differ from citations earned in ChatGPT, Claude, Perplexity, Gemini, or Google AI Mode. In a SearchVIU experiment, those systems extracted visible HTML during direct webpage retrieval. That test found no work with of JSON-LD, hidden Microdata, or hidden RDFa.
Why Visible Content Remains Central To AI Visibility Explained
My AI Optimization work begins with written material users may read. Clear answers, broad topic coverage, original evidence, and reliable entities give systems material they can interpret and trust.
Schema does not replace helpful writing. It cannot conceal weak explanations or offer facts absent from the site. I work with structured data to reinforce visible information while keeping the central answers plain, complete, and easy to follow.
The experiment does not define every role schema can play in crawling, indexing, training, or retrieval. For AI Optimization, I still prioritize site standard, useful structure, and evidence that stands on its own.
My AI SEO Strategy To Improve AI Citations
My AI Strategy begins with helpful written material, clear evidence, and close alignment with user intent. I study questions people ask on Google Search, Google AI Overviews, Google AI Mode, and conversational platforms in practice. Each site should remain easy to understand, verify, and use.
Create Content That AI Systems Can Understand And Trust
My AI SEO process answers specific questions early and explains complex ideas in plain language. I support each important claim with reliable sources, original research, practical examples, or direct experience. Clear author information support readers assess the expertise behind a page.
I keep high-utility pages accurate and current. Strong internal links connect related topics and resource users through a page copy cluster. This structure gives search engines and AI systems additional context for each subject.
- Apply direct answers before broad background information.
- Show real experience through examples, processes, and useful specifics.
- Examine facts, dates, statistics, and product information on a regular schedule.
- Organize pages with easy-to-follow headings and short, focused paragraphs.
Strengthen The Signals That Schema Cannot Replace: A Practical Guide
My AI SEO method covers technical work that protects access and page quality. I check crawlability, indexation, page speed, duplicate content, and mobile usability. A well-structured webpage cannot perform well when search engines cannot reach or method it.
I build a credible reputation through high-quality links and mentions from respected websites. I overview readers toward related pages with applicable internal links. I generate written material for the full search journey, from basic questions to detailed comparisons and purchase decisions.
Schema can clarify site details, but it cannot replace useful writing, expert knowledge, strong sources, or a trusted website. My AI Strategy treats structured data as support within a wider system. The central focus stays written material that serves people and gives AI systems clear, reliable information to interpret.
How I Would Test Schema SEO On An Individual Website: A Practical Guide
I would begin with 10 to 20 pages from one domain. Five to 10 pages would already have AI citations, while another five to 10 similar pages would serve as controls in practice. That design would establish a fair baseline for measuring AI Citation Growth.
Before adjustments, I would record citations from Google AI Overviews, Google AI Mode, and ChatGPT. I would add Schema Markup only to test pages in practice. Content, links, templates, and technical settings would remain unchanged throughout the test.
- Track every site for at least 30 days.
- Use a 60-day or 90-day window when possible.
- Compare test pages with control pages on each platform in practice.
- Measure whether the test group shows stronger AI Citation Growth.
I would compare the citation gap between the test and control groups in practice. I would not attribute a platform-wide change to Schema Markup without evidence in practice. A longer test was able to expose delayed effects that a 30-day window might miss.
The report would state the test’s primary limits. Schema types might be pooled together, while pages using JSON-LD may receive other changes simultaneously. This method would examine JSON-LD placed in the HTML.
Pages with zero citations are difficult to evaluate in many cases. No increase could mean Schema Markup had no effect. It could also mean those pages were unlikely to receive citations during the test. That distinction matters when measuring AI Citation Growth in real-world use.
Affordable SEO Services For Organic Search And AI Visibility: A Practical Guide
I supply affordable SEO services for businesses seeking stronger organic search rankings, qualified traffic, and lasting visibility. I am Anatoly Zadorozhnyy, an SEO expert with more than 18 years of experience helping companies adapt to updates in search and digital marketing.
My work has helped hundreds of businesses improve Google rankings and reach the first page for thousands of valuable search terms. I focus on steady growth instead of quick tactics that might lose benefit as search systems change.
I audit Schema Markup for AI Optimization, but I do not present it as a guaranteed path to AI citations in real-world use. My services also cover technical SEO, helpful written material, internal linking, search intent, and strategies supporting organic search rankings and AI visibility.
More information around my affordable SEO services is available at www.affordableseoexpert.com.
