Does schema still matter in AI search?

If you’ve spent time in SEO forums or reading SEO blogs, you’ve probably heard someone scream, “Use schema!” like it’s the magic key to all things AI. 

But here’s the reality: AI-driven tools like ChatGPT search don’t need schema to understand your content. 

Unlike traditional search engines that rely on structured data to “read” and interpret pages, they can understand plain text naturally, similar to how a human might.

With the rise of AI-generated answers in search, SEOs are scrambling to keep their content visible. 

One of the most debated questions is whether schema markup – the structured data SEOs have been told to prioritize for years – still matters. 

Do AI models actually need it to understand what’s on your page? Or is it an outdated SEO crutch?

The short answer:

• No, schema isn’t required for AI-driven search engines to understand your content.

The longer answer:

• Yes, it can still help – especially if you want to win featured snippets or have FAQ, how-to, or review content.

To make sure there is no misinterpretation of what I mean or my intent, I want to break down exactly when schema matters, when it doesn’t, and where I think you should focus your time instead. 

I’ll also explain how search engines like Google and Microsoft Bing assess authority and why it matters for visibility in ChatGPT search results.

How does ChatGPT pick which sites to cite?

Before discussing schema, it’s important to understand how ChatGPT search decides which sites to cite as sources. 

Unlike the classic (10 blue links) ranking system, ChatGPT search pulls in “the best answer,” not necessarily “the best page.”

Here are the four core principles that AI uses to decide what to cite:

1. Relevance first, authority second

When a user asks a question, the AI doesn’t just look for “the best page” – it looks for the best answer. This is a subtle but important shift.

If your page answers a user’s query directly and clearly, you have a shot at being included – even if you’re not the #1 organic result. But relevance alone isn’t enough. You also need to be seen as an authority on the topic.

Takeaway: If you want to be cited by ChatGPT search, write content that answers specific questions directly and make sure it clearly demonstrates authority (more on that below).

Dig deeper: A 13-point roadmap for thriving in the age of AI search

2. Authority = E-E-A-T

There is no scenario in the multiverse where you haven’t seen the term E-E-A-T (experience, expertise, authoritativeness, and trustworthiness) in every SEO conversation since 2022. 

If you’re saying to yourself, “Self, I’m pretty sure ChatGPT uses Bing, not Google, so why are we bothering with the E-E-A-T discussion?” 

Bing has a quality system similar to Google’s E-E-A-T, but it is called quality and credibility (QC). 

So yes, even though we’re talking about ChatGPT search and, therefore, Bing, the same concepts still apply.

Here’s why this matters:

• Both systems value authority, expertise, and trust. Call it E-E-A-T, call it QC – it’s the same fundamental idea.
• AI models like ChatGPT search rely on Bing’s index, so demonstrating credibility within Bing’s QC framework can improve your visibility in ChatGPT answers.

Unlike Google’s E-E-A-T, Bing’s QC is a little less publicized, but the principles are clear. 

For content to be seen as authoritative and credible, it needs to check certain boxes:

• Author bios that showcase expertise in the field.
• References and citations to credible external sources.
• Internal links that establish your site as a hub of knowledge.
• Content freshness (old content that hasn’t been updated often gets ignored).

Takeaway: Call it E-E-A-T or QC – the name doesn’t matter. Your content must demonstrate clear expertise and credibility if you want to rank in Google, Bing, and ChatGPT-driven search.

Dig deeper: Decoding Google’s E-E-A-T: A comprehensive guide to quality assessment signals

3. Consensus is king

Here’s something that might not be obvious if you’re thinking in “traditional SEO” terms: ChatGPT is looking for consensus. 

When AI generates an answer, it wants it to be “correct,” and one of the best ways for it to feel confident is to see the same information confirmed in multiple places.

But not all consensus is created equal.

• For hard facts (like the boiling point of water, title tag length, or the capital of France), AI will go with the clear, objective consensus.
• For subjective or “squishy” topics (like “Does pineapple belong on pizza?”), the AI will often try to present a more “balanced” view. If you’ve ever seen ChatGPT hedge its response with “Some people think X, while others believe Y,” this is why.

Example – hard fact:

• If every SEO site says title tags should be 50-60 characters, but you say they should be 97, guess which guidance AI is going to follow? (Hint: It’s not you.)

Example – squishy question:

• If Reddit says “pineapple on pizza is a crime,” but social media influencers start calling it “the flavor combo of the future,” guess what happens? AI won’t pick a side – it will land on “some people love it, others hate it.”

Takeaway: For factual questions, align with the broader consensus. For opinion-based questions, being a dissenting voice can shape the AI narrative.

Dig deeper: Exploit reveals how and why Google ranks content

4. The ‘credibility layer’ (Yes, it’s a thing)

AI-driven models like ChatGPT search are believed to use a “credibility layer” when deciding which sources to trust. 

Similar to how spam filters evaluate the trustworthiness of email senders, this system goes beyond just “what’s on the page” and looks at broader signals. 

Domain history, link quality, brand reputation, and user behavior are all factors that could be part of this credibility check.

Here’s what might be in the “credibility layer” (based on patents, research, and common sense):

• Domain history: New sites have a more challenging time getting cited.
• Inbound link quality: Not volume – quality.
• Brand reputation: Are you mentioned on trusted sites?
• User behavior: If people frequently “return to search” after visiting your page, that’s not a good signal.

Takeaway: Instant authority isn’t a thing. Build it over time.

Does schema matter for ChatGPT search?

Large language models (LLMs) like ChatGPT don’t need schema because they can “read” natural language and understand it in context. 

They know that a list of ingredients followed by a set of instructions is a recipe. They recognize headings, subheadings, and lists – no schema required.

But here’s the nuance:

Just because they don’t “need” it doesn’t mean it’s useless. 

Schema still helps by making it easier for AI to identify key parts of your content. 

It’s like wearing a name tag at a conference – people could still figure out who you are, but it’s faster and easier if you just wear the tag.

Dig deeper: SEO for ChatGPT search: 4 key observations

The SEO playbook: What should SEOs focus on instead of schema?

If you’ve made it this far, you already know the answer: relevance, authority, consensus, and credibility are the real factors. 

This isn’t just a “Google thing” – these are the same principles used by Bing (QC), Google (E-E-A-T), and AI-driven search engines like ChatGPT search.

Here’s your SEO playbook for AI-driven search:

• Relevance: Answer specific questions directly.
• Authority: Show clear expertise with bios, fresh content, and links to credible sources.
• Consensus: If your stance on a topic is wildly different from everyone else’s, you’d better have proof.
• Credibility: Build your reputation, brand mentions, and link profile over time.

Want to future-proof your SEO strategy? Focus on being the most trusted answer on the web. E-E-A-T and QC are just labels – the real job is to show your work.

TL;DR: Should you use schema for ChatGPT search?

No, it’s not required. But yes, it can help.

• Use schema for FAQs, how-to guides, reviews, and product pages.
• Don’t waste time on listicles, blog archives, or service pages.
• If you’re serious about staying ahead in SEO, focus on relevance, authority, consensus, and credibility.

The era of “blindly adding schema to everything” is over. Use it strategically where it matters.

For everything else, focus on being clear, useful, and undeniably credible.

Dig deeper: AI SEO: How to be visible in Google AI Overviews, chatbots, LLMs