No, adding schema markup alone does not automatically improve AI citations in Google AI Overviews, ChatGPT, or AI Mode.
A recent Ahrefs study tracking 1,885 pages found little to no measurable improvement after adding JSON-LD schema. For local businesses, stronger visibility still appears to come from trust, authority, consistent business signals, review ecosystems, and topical relevance rather than simply adding more structured data.
Why “Schema for AI Citations” Became a Big SEO Conversation
Over the last year, AI search changed how many businesses think about SEO.
Instead of asking:
- “How do we rank higher?”
Businesses started asking:
- “How do we get cited by AI systems?”
That shift created an entirely new layer of SEO discussions around:
- AI retrieval
- entity understanding
- AI summaries
- Google AI Overviews
- conversational search
- answer engines
And somewhere in that discussion, schema markup became positioned as a supposed shortcut for AI visibility.
The argument usually sounded like this:
“AI systems need structured data. Add more schema and your business becomes easier for AI to understand.”
At surface level, that sounds reasonable.
But the actual evidence is far more nuanced.
What Ahrefs Actually Tested
Ahrefs conducted one of the most discussed SEO studies of 2026 around schema for AI citations.
They tracked 1,885 URLs that added JSON-LD schema between 2025 and 2026.
Then they compared those pages against similar control pages that did not add schema.
The question was simple:
Did schema markup increase citations inside:
- Google AI Overviews
- Google AI Mode
- ChatGPT
The findings surprised a lot of SEOs.
Results:
- Google AI Mode → almost no meaningful change
- ChatGPT → almost no measurable improvement
- Google AI Overviews → slight decline relative to control pages
That does not mean schema hurts rankings or visibility.
But it does challenge the idea that schema alone is an AI citation growth strategy.
The Important Context Many People Missed
This detail changes the entire interpretation of the study.
The pages Ahrefs tested were already heavily cited by AI systems before schema was added.
That means those pages were already:
- indexed
- crawled
- trusted
- retrieved
- visible in AI ecosystems
So the study was not testing:
“Can schema help pages become visible?”
It was testing:
“Can schema increase visibility for pages already visible?”
Those are completely different scenarios.
This is where many simplified LinkedIn opinions became misleading.
Why This Matters for Local SEO
Most local businesses are not competing for national AI citations.
A dentist in Cork is not trying to become a globally cited AI entity.
A roofing company in Dublin is not competing with Wikipedia.
Local businesses are competing for:
- local intent visibility
- Google Maps prominence
- branded searches
- service authority
- trust signals
- regional relevance
That changes how schema should be prioritized.
What Local Businesses Actually Need More Than Schema
Many businesses now believe advanced schema implementations are the key to AI visibility.
But most local businesses still have foundational SEO problems.
The businesses winning local visibility usually have:
- strong Google Business Profile optimization
- accurate NAP consistency
- localized service content
- strong review ecosystems
- crawlable websites
- location authority
- branded search behavior
- topical coverage
- established trust signals
Schema can support these systems.
But it does not replace them.
Why Correlation Confuses the SEO Industry
One of the most valuable insights from the Ahrefs study is not actually about schema itself.
It is about how SEO myths form.
Ahrefs observed that pages with schema were often:
- more authoritative
- technically stronger
- better maintained
- more established
- more linked
So naturally, those pages were also more visible.
That creates correlation.
But correlation is not proof of causation.
This mistake happens repeatedly in SEO:
- A signal appears alongside successful sites
- The industry treats it as a ranking factor
- Agencies package it into services
- Businesses overspend on tactical implementation
- The actual visibility drivers get ignored
Schema for AI citations is now entering that cycle.
AI Systems Probably Prioritize Trust More Than Markup
One pattern becoming increasingly visible in AI search is retrieval confidence.
AI systems appear more likely to surface businesses that demonstrate:
- authority
- historical consistency
- entity clarity
- review trust
- brand mentions
- content depth
- real-world prominence
Older and more established businesses naturally accumulate these signals over time.
That matters because AI systems do not just retrieve pages.
They retrieve confidence.
The Role of Entity Understanding in Local SEO
Entity SEO is becoming increasingly important in AI search environments.
Google and AI systems attempt to understand:
- who a business is
- where it operates
- what services it provides
- how consistently it appears across the web
- whether other sources validate it
This is where schema still has value.
Schema can help reinforce:
- organization identity
- local business information
- authorship
- service relationships
- knowledge graph connections
But again:
schema supports understanding.
It does not create authority by itself.
Google Quietly Reduced the Impact of Some Schema Features
Another overlooked reality is that Google has already reduced visibility for some schema-powered features.
FAQ schema is a major example.
In 2026, Google confirmed broader FAQ rich results were being deprecated for most websites.
That is important.
Because for years, many SEO campaigns treated FAQ schema as a guaranteed visibility tactic.
Now Google is becoming far more selective.
This signals a broader trend:
- structured data still matters
- but visible SERP enhancements are becoming more restricted
Where Schema Still Makes Sense
Schema markup still provides value in many situations.
Especially for:
- Local Business schema
- Product schema
- Organization schema
- Event schema
- Review-related markup
- Merchant information
- Knowledge graph support
For ecommerce and multi-location brands, schema remains operationally useful.
The mistake is assuming:
“More schema automatically means more AI visibility.”
That conclusion is not supported by current evidence.
What I Would Prioritize First for Local Businesses
Before investing heavily into advanced schema projects, I would focus on:
1. Google Business Profile Optimization
GBP remains one of the strongest local visibility systems in 2026.
2. High-Quality Local Landing Pages
Thin location pages continue to struggle with trust and AI retrieval.
3. Entity Consistency
Consistent business information across platforms matters significantly.
4. Topical Authority
Businesses deeply covering their services build stronger trust signals.
5. Review Ecosystems
Reviews still influence both human trust and local search visibility.
FAQ
Does schema markup help AI citations?
Current evidence suggests schema alone does not significantly improve AI citations for pages already visible in AI systems.
Is schema markup still important for SEO?
Yes. Schema still supports search understanding, rich results, entity clarity, and structured interpretation.
What matters more than schema for local SEO?
Trust signals, GBP optimization, review quality, topical authority, entity consistency, and local relevance.
Should local businesses stop using schema?
No. Schema still provides value, but it should support a broader SEO strategy instead of replacing one.
Final Thoughts
The Ahrefs study forced the SEO industry to confront an uncomfortable but important reality:
Not every SEO correlation is a ranking factor.
Schema markup still has legitimate technical and semantic value.
But local businesses should avoid treating schema as a standalone AI visibility strategy.
The businesses likely to win AI-driven search visibility in the coming years are not simply the businesses with the most markup.
They are the businesses with the strongest trust ecosystems.
If you want help improving your Local SEO, GBP visibility, or AI search readiness, visit: