Key Points
- Ahrefs took 1,885 pages that added JSON-LD schema between August 2025 and March 2026 and compared each one with control pages from other sites that had similar citation levels and never added schema.
- The limits matter, because they tell you where schema might still help.
- None of this makes schema useless.
- From the research we have read and from tracking AI answers for our own clients in the Carril portal, these are the things worth your time first.
Adding schema markup to your pages will not, on its own, get your brand cited by ChatGPT or Google’s AI answers. The largest controlled study so far, published by Ahrefs in May 2026, tracked 1,885 pages that added schema and found no clear lift in AI citations. Schema is still worth having for other reasons, but if your plan for AI visibility starts and ends with structured data, it will not move much.
This post covers what the study found, what it could not test, what schema still does well, and what does change whether an AI assistant names you.
What the Ahrefs study found
Ahrefs took 1,885 pages that added JSON-LD schema between August 2025 and March 2026 and compared each one with control pages from other sites that had similar citation levels and never added schema. It counted AI citations for 30 days before and 30 days after each page changed, and tested the difference several ways.
| Platform | Change in citations after adding schema | What it means |
|---|---|---|
| ChatGPT | +2.2% | No real difference from pages that added nothing |
| Google AI Mode | +2.4% | No real difference |
| Google AI Overviews | -4.6% | A small drop, which Ahrefs could not tie to the schema |
In the author’s words, “not much really changed: schema had no clear positive or negative effect.” Her conclusion for teams that already do the rest of their SEO well: “JSON-LD isn’t going to be the unlock.”
What the study could not test
The limits matter, because they tell you where schema might still help.
- Every page was already visible to AI. The sample only included pages with more than 100 AI Overview citations. The study says nothing about a page that AI tools do not cite yet.
- The window was 30 days. Slower effects would not show.
- All schema types were pooled together. Article, FAQ, Product, HowTo and Organization markup were counted as one.
- Pages often changed other things at the same time, which blurs the effect of the schema itself.
- Only JSON-LD in the page’s HTML was tested, not markup added later by JavaScript.
So the honest reading is narrow: for pages that AI tools already cite, adding schema does not raise how often they are cited.
What schema still does well
None of this makes schema useless. It still does the jobs it was built for:
- Rich results in Google. Product prices and stock, review stars, events and breadcrumbs can still show in normal search results when the markup is right.
- Telling search engines what a thing is. Organization and LocalBusiness markup tie your name, address, logo and social profiles together, which helps your business be understood as one entity.
- Keeping facts consistent. A well-kept schema block is one place where your opening hours, prices and contact details are written the same way every time.
We still add schema to the sites we build. We just do not sell it as an AI visibility tactic.
What does change whether an AI assistant names you
From the research we have read and from tracking AI answers for our own clients in the Carril portal, these are the things worth your time first.
Answer the question in the first lines of the page. Assistants quote passages that answer a question directly. A page that opens with two paragraphs of scene-setting gives them nothing to lift.
Make your pages readable without JavaScript. Many AI crawlers do not run JavaScript. If your content, or your prices, only appear after a script runs, those crawlers see an empty page. Check what your page looks like with JavaScript turned off.
Describe your business the same way everywhere. Your website, Google Business Profile, directory listings and social profiles should agree on what you do, where you are and who you serve. Assistants trust what they see repeated.
Be mentioned on other sites. AI answers lean on pages that already rank and on sources that name you. Reviews, directory listings, press coverage and appearances in list articles all count.
Measure it with a fixed set of questions. Write down the questions your buyers ask before they choose a business like yours, ask them in ChatGPT, Perplexity, Gemini and Google on a schedule, and record whether you are named and which pages are cited. That is how we track it for clients, and it is the only way to know whether any change worked.
Where to start this week
- Pick your three most important pages and read their first two sentences. If they do not answer the question the page targets, rewrite them.
- Open each page with JavaScript turned off in your browser. If the main content disappears, that is your first fix.
- Ask ten buyer questions in ChatGPT and Perplexity and write down who is named. That is your baseline.
- Keep your schema, check it is valid in Google’s Rich Results Test, and move on.
If you would rather have this done for you, our GEO and SEO service covers the site changes and the tracking, and you can check one page now with our free AEO score tool.
Sources
- Ahrefs, Louise Linehan, “We Tracked 1,885 Pages Adding Schema. AI Citations Barely Moved”, 11 May 2026: ahrefs.com/blog/schema-ai-citations
- CMSWire, “Does structured data actually improve answer engine citations?”: cmswire.com
Frequently Asked Questions
Does schema markup help with AI search?
Not much, for pages that AI tools already cite. Ahrefs found that adding JSON-LD made no clear difference to citations in ChatGPT or Google AI Mode. It may still help pages that are not cited yet, which the study could not test.
Should I remove my schema markup?
No. Schema still earns rich results in Google, ties your business details together and costs little to keep. It just should not be your plan for AI visibility.
Does FAQ schema help ChatGPT cite my page?
There is no controlled study showing that it does. The Ahrefs study pooled FAQ markup with other types and found no lift. Writing clear questions and answers on the page itself matters more than the markup around them.
What helps a page get cited by AI assistants?
Direct answers near the top of the page, content that loads without JavaScript, consistent facts about your business across the web, and mentions on other sites that AI tools already trust.
How do I know if my brand appears in AI answers?
Ask a fixed set of buyer questions across the main assistants every few weeks and record whether you are named and which pages are cited. Tools exist to automate this, and the Carril portal runs it for our clients.
