fbpx

Why SEOs Won’t Stop Talking About Structured Data (And You Should Too!)

3 min read

A recent LinkedIn post questioned whether Schema.org structured data influences the output of large language models (LLMs). Despite skepticism, some SEOs recommend using structured data as a strategy to rank better in AI-powered search engines. But does this claim hold water?

 

Fact vs. Opinion in SEO

 

This discussion highlights a recurring issue in SEO: the blurred line between fact and opinion. Opinions are valuable—they spark ideas and innovation—but problems arise when opinions are treated as facts. For example, when someone says, “I think,” it signals an opinion. However, these opinions can snowball into widely accepted “truths” without solid evidence.

 

The LinkedIn Post That Sparked the Debate

 

SEO expert Patrick Stox raised an important question:

“Did I miss something? Why do SEOs think schema markup will impact LLM output?”

His question refers to LLMs like those behind ChatGPT Search or other AI-driven search engines. Stox’s inquiry prompts us to ask: do AI search engines actually rely on Schema.org structured data to generate results?

 

How LLMs and AI Search Engines Work

 

LLMs like ChatGPT and Bard are trained on vast datasets, including web content, books, and official documents. While they produce summaries and responses, they don’t directly send traffic to websites or rely on structured data for training.

AI search engines use retrieval-augmented generation (RAG) methods, which combine a search index and knowledge graphs to generate results. These indexes are built from crawled web content, not Schema.org data. For instance:

  • Google and Bing: Use crawled HTML content for indexing, rendering pages to extract headers, main content, and other text—not relying solely on structured data.
  • Perplexity AI: Uses a modified version of PageRank to rank web-crawled content.

Structured data is primarily used by Google for specific rich results, such as FAQs or product snippets, and even then, it utilizes only a fraction of the available Schema.org vocabulary.

 

Misinterpretations and the “Game of Telephone” Effect

 

The notion that Schema.org structured data can improve AI search rankings seems rooted in misinterpretation. For example, Jono Alderson once suggested that structured data could help AI search engines better understand web content. However, this idea may have evolved through repetition into an unsupported SEO “theory.”

SEO is no stranger to unsubstantiated claims. For example, some SEOs argue that Google’s Local Search doesn’t use IP addresses for “near me” queries—a claim easily debunked by testing with a VPN. Google even documents that IP addresses influence local results.

 

Schema.org and AI Search: A Mismatch

 

Believing that Schema.org structured data impacts LLM output is flawed because:

  1. LLMs Don’t Cite Training Data: LLMs generate responses based on training data but don’t source structured data directly.
  2. Search Engines Focus on Text, Not Schema: Search engines primarily use HTML content for indexing and ranking.

Even within structured data, only a small subset is used for search features like rich snippets. AI search engines, therefore, have little reason to prioritize Schema.org data.

 

Pragmatic SEO: A Call for Evidence-Based Practices

 

Christopher Shin, commenting on Stox’s post, added a nuanced perspective:

“LLMs typically don’t generate responses from search engine SERPs but rather from data interpretation. Schema markup is used for SERPs, not LLMs.”

This distinction underscores the importance of pragmatic SEO—focusing on realistic, evidence-based strategies rather than speculation.

 

Closing Thoughts

 

SEO has a history of chasing myths and wasting time on vague theories. Misunderstandings, like those about structured data and AI search, only add noise to an already complex field. By prioritizing clarity and critical thinking, professionals like Patrick Stox and Christopher Shin remind us of the value of sensible, informed SEO practices.

Let’s not fall into the trap of conflating opinion with fact. Instead, let’s strive for strategies grounded in reality, supported by data, and aligned with how search engines—and now AI—actually work.

If you’re still feeling overwhelmed, don’t worry—our monthly SEO packages are here to make things easy. Let the experts handle it for you!

Shilpi Mathur
navyya.shilpi@gmail.com