Schema Markup and AI Citations: What the 2026 Data Actually Says
A controlled 2026 Ahrefs study found schema markup doesn't lift AI citations. Here's what the data actually shows and why to build it anyway.
In February 2024, Gartner predicted that by 2026, traditional search engine volume would drop 25% as AI chatbots become substitute answer engines (source). That’s this year. Schema markup is the answer most agencies will sell you first for AI search visibility — here’s what the evidence actually says.
Ahrefs tracked 1,885 pages that added JSON-LD schema between August 2025 and March 2026, compared them against 4,000 control pages, and measured citation changes. The result: essentially no lift. AI Mode moved +2.4%, ChatGPT +2.2% — both statistically meaningless — and AI Overviews declined 4.6% (source).
Why the stat everyone quotes is misleading you
65-71% of pages cited by AI tools do carry structured data (source). But that’s correlation: pages that bother to add schema also tend to already be doing the deeper technical SEO, content depth, and link-building that actually earns the citation (source). As Ahrefs put it: if you’re already doing the rest of the SEO work well, JSON-LD isn’t going to be the unlock (source).
What AI tools actually “see” when they visit your site
A searchVIU test across ChatGPT, Claude, Perplexity, Gemini, and Google’s AI Mode found none of five systems could extract a price that only existed in hidden schema code (source). OpenAI’s crawler skips roughly 1 in 9 JavaScript-heavy pages, and Anthropic’s closer to 1 in 4.
So should you spend money on this or not?
Keep doing basic schema as routine technical hygiene, but stop paying a premium for it as an “AI SEO” silver bullet. Three real reasons to still have it: it’s foundational hygiene for entity disambiguation, it’s cheap once automated, and it supports measurement — tracking which pages get cited and correlating that against content and distribution work.
What to tell an agency pitching “AI schema optimization”
Ask them directly how that claim holds up against the Ahrefs controlled study. A credible answer repositions schema as hygiene, not a growth lever.
FAQ
Should we rip out our existing schema markup? No — keep it as hygiene.
Is paying extra for “AI schema optimization” worth it? Generally not as a standalone service.
Did Google’s recent changes affect this? Somewhat — Google removed FAQ rich results from Search shortly before this study came out.
What should we spend AI-search budget on instead? Deeper content, technical SEO fundamentals, and earned links and mentions.
How do we check our AI crawler access? A raw HTML check (no JavaScript rendering) shows what a JS-skipping crawler actually sees.
See this in production at /workflows/schema-structured-data-ai-citations. If your team needs an evidence-based AI search strategy, that’s covered at /answer-engine-optimization.