Does Schema Data Really Matter in AI Search? Here Is What the 2026 Research Says


Updated June 24, 2026

Image

Written by

Hassam



Here is what the 2026 research actually says. And here is why the answer is more complicated than the people arguing loudest on LinkedIn would have you believe.

Every few weeks, the SEO community divides itself into two camps over schema markup. Camp one says structured data is critical for AI citation and you are invisible without it. Camp two says LLMs flatten everything to plain text so schema is irrelevant.

Both camps are partially right. Neither camp is telling you the full story.

We looked at the actual 2026 research. Here is what it says, what it means for your site, and what you should actually be prioritising right now.

The Debate, and Why It Exists

The “schema is dead for AI” argument has a logical foundation. When a large language model processes a page, it tokenises the content. Structured data written in JSON-LD gets converted to plain text in that process, which means the hierarchical structure you carefully built disappears before the model reasons with it.

The counter-argument is equally logical: even if structure dissolves at the token level, schema still signals entity relationships, content type, and factual claims during the crawling and indexing phase, before the model ever sees the page.

Both arguments describe something real. The problem is that neither argument is based on controlled experimental data. Until recently, nobody had actually tested it at scale.

The schema debate has been argued almost entirely without data. The SEO community has been split on this question for 18 months based largely on opinion, not measurement. That changed in May 2026.

What the Largest Study Found

In May 2026, Ahrefs published the most rigorous schema markup study conducted in the AI search era. They tracked 1,885 pages that added JSON-LD schema between August 2025 and March 2026, matched against 4,000 control pages with similar pre-existing citation levels. They then measured AI citation changes 30 days before and 30 days after the schema was added.

Adding schema markup produced near-zero measurable increase in AI citations across Google AI Mode and ChatGPT. The figures sat close enough to zero to be classified as random noise across thousands of URLs.
Source: Ahrefs, 1,885-page schema study, May 2026

That is a significant finding. It directly contradicts the loudest pro-schema voices in the GEO community.

But before you uninstall your schema plugin, read the caveat Ahrefs included in the same study: every page in their dataset was already being heavily cited by AI systems, with 100 or more AI Overview citations before any schema was added. For pages already in the citation pool, adding more schema did not move the needle.

For pages not yet in the citation pool, the study could not draw conclusions. That gap matters.

What this means practically
Schema is not a ticket into AI citations. If your content, brand authority, and topical relevance are not there yet, adding JSON-LD will not change your AI visibility. Schema is not the entry barrier. But it may be part of the infrastructure that helps you stay credible once you are in the game.

But the Other Data Is Not All Noise Either

Here is where it gets genuinely complicated. Multiple 2026 studies point in a different direction to the Ahrefs findings.

Schema markup is the cheapest 2.3x citation lift in an April 2026 dataset of 1,000 AI Overviews analysed across 4,200 unique cited URLs.
Source: Digital Applied, 1,000 AI Overviews Citation Pattern Study, April 2026

A separate OtterlyAI experiment, running from November 2025 through March 2026 across 319 tracked prompts, found that Google AI Mode and AI Overviews were the two platforms that showed a measurable increase in AI visibility after schema implementation. The AI Overviews increase was, in their words, “substantial.”

A March 2026 analysis from Digital Applied found that the shift in Google’s schema processing was moving from schema as a rich result trigger to schema as an entity trust and verification signal. In other words, schema is now less about getting a star rating in search results and more about helping Google confirm that your organisation is a real, credible entity.

The single highest-leverage schema implementation in 2026 is entity markup that identifies your organisation in Google's Knowledge Graph. It produces no visible SERP feature, but its impact on AI Mode citation and Knowledge Panel accuracy is measurable.
Source: Digital Applied, Schema Markup After March 2026

So which is it? Schema matters or it does not?

The honest answer: schema as a direct citation trigger for LLMs like ChatGPT is weak. Schema as an entity signal for Google’s own AI infrastructure, specifically AI Mode and AI Overviews, carries measurable weight. These are different systems with different architectures, and treating them as the same thing is the source of most of the confusion.

The Bigger Story Behind the Schema Debate

Fixating on schema markup is a symptom of a deeper problem. Most UK businesses are still optimising for a search model that no longer describes how their buyers find them.

The model most SEO strategies are still built on: create a page, rank it in Google, generate clicks, convert visitors.

The model that is actually operating in 2026:

93% of AI search sessions end without a website click. AI Overviews alone now trigger on 25% of all Google searches, up from 13% in March 2025.
Source: Superlines AI Search Statistics, 2026 / Conductor analysis of 21.9 million queries

Clicks are no longer the primary unit of value in search. Mentions are.

When a buyer asks ChatGPT which digital marketing agency to use in the UK, they are not seeing a list of links. They are receiving a synthesised recommendation. Your brand either appears in that recommendation or it does not. No click is involved. But the buying decision is influenced.

At the product discovery stage, 35% of US consumers now use AI tools versus 13.6% who use search engines. At the evaluation stage, AI holds a 32.9% to 15% advantage over search.
Source: Similarweb 2026 Generative AI Brand Visibility Index, January 2026

That is not a future trend. That is the current buying behaviour of your prospects.

The AI conversation three days before purchase is setting the shortlist. The Google search that precedes the final click is closing a decision that was already mostly made. Under last-click attribution, Google gets the credit. The AI mention made the sale possible and went unmeasured.

90% Of Brands Have Zero AI Search Mentions. They are visible in the old game and invisible in the new one. 
Source: SEJ 2026 GEO Visibility Research.

What Actually Drives AI Citation in 2026

We have now established that schema is a supporting signal, not a primary driver. So what are the primary drivers?

The honest answer comes from several large-scale 2026 studies and points to the same five signals:

  1. Branded Web Mentions Across Third-Party Sources

Ahrefs’ analysis of 75,000 brands found branded web mentions correlate with AI visibility at 0.66 to 0.71. The AirOps 2026 State of AI Search report found that 85% of brand mentions originate from third-party pages, not owned domains. Brands are 6.5 times more likely to be cited in AI answers through third-party sources than through their own websites.

This is PR and digital authority work, not technical SEO. You need to be talked about by publications, communities, and platforms that AI systems already trust.

  1. Content Structure, Not Just Content Volume

SE Ranking’s analysis found that pages structured into sections of 120 to 180 words earn 70% more ChatGPT citations than pages with short dense paragraphs. Pages with FAQ sections correlate with significantly higher citation rates across platforms. Content with statistics, citations, and quotations achieves 30 to 40% higher visibility in AI responses.

This is the GEO discipline of writing for extractability, not just readability.

  1. Content Freshness

Pages updated within the past two months are twice as likely to be cited as older pages. In fast-moving industries like SaaS, fintech, and B2B services, stale content is invisible to AI systems that are actively pulling fresh web data to construct answers.

  1. YouTube Mentions

This is the signal that most SEO strategies completely ignore. Ahrefs’ brand study found YouTube mentions show a 0.737 correlation with AI visibility, the strongest of any single signal measured. Brands mentioned in YouTube video titles, transcripts, and descriptions appear more consistently across AI platforms than brands with equivalent web authority but no YouTube presence.

  1. Dual Signal Presence: Citation and Mention
Brands that earn both a mention and a citation in AI answers are 40% more likely to resurface across consecutive queries than citation-only brands. Only 30% of brands stay visible across back-to-back AI answers.
Source: AirOps 2026 State of AI Search Report

Most AI visibility strategies chase citation. Fewer invest in the brand mention layer, which is the signal that keeps you consistently surfaced. Both signals require different work. Citations come from well-structured, authoritative content. Mentions come from consistent brand presence across communities, platforms, and earned media.

The HDSQUARES Position

HDSQUARES Verdict

Schema is worth implementing correctly, specifically entity markup and Organisation schema for Knowledge Graph clarity. But it is maintenance work, not growth work. If your AI visibility strategy starts and ends with structured data, you are optimising a supporting signal while ignoring the primary drivers. The real work is brand presence, content extractability, and earned mentions across third-party platforms.

This is exactly why we separated GEO from SEO as a distinct service. They share infrastructure. They do not share strategy.

A well-maintained technical SEO foundation, including valid schema, gives AI systems the cleanest possible read of your content. But the brands consistently appearing in AI recommendations for their category are not there because of their JSON-LD. They are there because:

  • They are mentioned consistently across trusted publications, communities, and editorial sources
  • Their content answers real buyer questions in a format that AI systems can extract and cite directly
  • Their brand is present across YouTube, LinkedIn, and industry forums where AI systems increasingly draw training and retrieval data
  • They have built topical authority across a specific domain rather than producing broad, shallow content across many topics

We have applied this framework to HDSQUARES own products. My Holiday Map, which we built and launched, generated over 20,000 euros in revenue. The strategy was not centred on technical optimisation. It was centred on brand presence, multi-channel distribution, and content structured for the way buyers actually discover and decide.

The same principles apply to every client we work with in B2B services, SaaS, and professional services. The businesses winning AI visibility in 2026 are not winning it by fixing their schema. They are winning it by building brands that AI systems consistently encounter across the web.

Where HDSQUARES can help
We audit your current AI visibility across ChatGPT, Perplexity, Google AI Mode, and Gemini. We identify the gap between where you rank on Google and where you appear in AI answers. And we build a GEO strategy that addresses the actual citation drivers, not just the technical layer. If this is a gap you are aware of but have not yet addressed, that conversation starts at hdsquares.com.

What You Should Actually Do With Your Schema Right Now

Do not delete it. Do not obsess over it. Do this:

  • Implement Organisation schema with a stable @id that links to your Knowledge Graph entity. This is the highest-value schema investment in 2026.
  • Add FAQ schema to service pages and blog posts, but only where the questions are real buyer questions with direct answers. Generic FAQs add noise.
  • Validate everything. Invalid schema is actively unhelpful. A broken FAQ schema block is worse than no schema at all.
  • For e-commerce and service businesses, use BreadcrumbList to reinforce site structure. This helps AI systems understand your content hierarchy.
  • Stop treating schema as your primary GEO strategy. Allocate more time to the signals that actually drive citation: content structure, freshness, third-party mentions, and YouTube presence.

Frequently Asked Questions

Does schema markup help you appear in Google AI Overviews?

The evidence is mixed. Ahrefs’ May 2026 study of 1,885 pages found that adding schema produced near-zero citation lift for pages already in AI systems’ consideration set. However, a separate April 2026 analysis of 1,000 AI Overviews found schema markup to be the cheapest 2.3x citation lift in the dataset. The most useful schema investment for AI Overviews specifically is entity markup that verifies your organisation in Google’s Knowledge Graph, rather than content-level schema types.

Does schema data help ChatGPT or Perplexity cite your content?

The evidence here is weaker. LLMs flatten structured data to plain text during token processing, which means the hierarchical structure of JSON-LD does not survive the way it does in Google’s crawling and indexing phase. For ChatGPT and Perplexity, content structure, freshness, branded web mentions, and third-party citations are stronger predictors of citation than schema type or completeness.

What is the most important thing for AI search visibility in 2026?

Branded web mentions and third-party editorial presence. Ahrefs’ analysis of 75,000 brands found that brands in the top 25% for web mentions earn over 10 times more AI Overview citations than those in the next quartile. No technical SEO signal comes close to that correlation. Building consistent brand presence across publications, communities, and trusted platforms is the highest-leverage investment for AI visibility.

What is GEO and how is it different from SEO?

Generative Engine Optimisation is the practice of structuring your content and brand presence so that AI platforms like ChatGPT, Perplexity, and Google AI Mode cite and recommend you. Traditional SEO targets position in ranked link lists. GEO targets inclusion in AI-generated answers, which may produce no direct click but significantly influences brand trust and downstream purchase behaviour. The signals that drive GEO performance are different from those that drive organic rankings.

Should I remove schema from my website?

No. Valid schema still supports traditional rich results in Google, improves Knowledge Graph entity clarity, and may contribute to AI Mode citation at the entity level. The point is not to remove it. The point is to stop treating it as a primary GEO strategy and invest appropriately in the signals that actually drive AI citation: content structure, freshness, and earned brand presence.