Does Structured Data Help AI Visibility? New Study Reveals Schema Does Not Boost AI Citations
The Quest for AI Visibility: Does Schema Actually Work?
For years, search engine optimization (SEO) specialists have leaned on schema markup—structured data that helps search engines understand the context of a page—as a primary tool for enhancing visibility. However, as the digital landscape shifts toward Generative AI and Large Language Models (LLMs), a critical question has emerged: Does structured data actually influence whether an AI cites your content in its responses?
A comprehensive new study conducted by Ahrefs suggests that the answer is a resounding ‘no.’ The research indicates that adding JSON-LD schema does not provide a statistically significant boost in citations across the major AI platforms, including Google’s AI Overviews and OpenAI’s ChatGPT.
Detailed Methodology of the Ahrefs Study
To reach these conclusions, Ahrefs implemented a rigorous testing framework designed to isolate the impact of structured data. Between August 2025 and March 2026, the team tracked 1,885 unique web pages that had schema added to them. To ensure the accuracy of the data, these ‘treated’ pages were matched against a control group of 4,000 similar pages that remained without the added schema.
The researchers measured citation changes across three primary AI-driven environments:
- Google AI Overviews: The generative AI summaries appearing at the top of Google Search results.
- Google AI Mode: The dedicated conversational AI interface.
- ChatGPT: OpenAI’s widely used chatbot and its integrated web-browsing capabilities.
The Results: No ‘Magic Bullet’ for AI Citations
The findings were surprising to many in the SEO community. Ahrefs reported that adding schema produced no major uplift in citations on any of the tested platforms. In some cases, the results were unexpectedly negative.
The Google Impact: In the case of Google AI Overviews, the study actually observed a 4.6% decline in citations for pages with schema compared to the control group. While this decrease is small, the researchers noted it was statistically significant, suggesting that schema is certainly not a catalyst for growth in this area.
ChatGPT and AI Mode: While pages with schema performed slightly better on average within ChatGPT and Google’s AI Mode, the researchers cautioned that these differences were minimal. The data suggests these fluctuations are likely ‘random noise’ across thousands of URLs rather than a direct result of the structured data implementation.
What This Means for the Future of GEO and SEO
This study has sent ripples through the Generative Engine Optimization (GEO) and SEO communities. For years, the industry has operated under the assumption that making data ‘easier for machines to read’ would naturally lead to more citations. These results suggest that AI models may be relying more heavily on the raw textual content, semantic relevance, and authority of a page rather than the structured tags wrapped around it.
While schema remains valuable for traditional rich snippets and enhancing a site’s overall technical health, it appears that it is not the key to unlocking more citations in the age of AI. Marketers may need to pivot their focus toward deeper content quality and user intent optimization to secure a spot in AI-generated answers.