Is Schema Markup Becoming the New Meta Keywords Tag?

There is a pattern in SEO that repeats itself every few years. Someone discovers a technical signal, the industry inflates its importance beyond recognition, and then a wave of agencies starts selling that signal as a premium service, charging clients handsomely for something that barely moves the needle. We saw it with meta keywords. We are watching the exact same thing happen right now with schema markup.

I have been doing SEO long enough to recognize when a concept gets hijacked by people who need something tangible to sell. Schema markup is genuinely useful in specific, well-defined scenarios. But what is happening in the market today, especially driven by the fake “AI optimization” crowd, is a distortion of reality that is costing business owners real money for work that produces almost no measurable benefit.

What Schema Markup Actually Does (And What It Was Always Meant For)

Schema markup is a form of structured data vocabulary from Schema.org that helps search engines understand the context of specific content elements on a webpage. Its primary and most practical purpose is to enable rich snippets in Google search results, such as star ratings, FAQ dropdowns, recipe cards, and event listings. It was never intended to be a broad ranking signal or an AI optimization tool.

Schema markup, when implemented correctly, can produce rich results in Google SERPs. That is its job. A local business might add LocalBusiness schema. An e-commerce site might add Product and Review schema. A recipe blog adds Recipe schema. The purpose is narrow but legitimate: give Google structured signals so it can display your content in a visually enhanced format.

That is it. That is the actual use case.

What schema was never designed to do:

• Directly improve organic rankings in standard blue-link results
• Make your website more “AI-friendly”
• Help large language models understand your content better
• Increase your chances of being cited in AI-generated answers
• Replace the need for strong content, authoritative backlinks, or topical depth

These distinctions matter enormously, because the current market is blurring all of them.

Nobody Was Talking About Schema for AI Until the “AIO” Industry Needed Something to Sell

Let me be blunt about what happened here. A new wave of “AI optimization” companies emerged, many of them repurposed SEO agencies with a fresh coat of paint, and they needed a concept they could sell with some visible proof of work. The challenge with convincing clients their content strategy improved was that content improvements are slow, competitive, and hard to attribute. Backlinks take time. Technical SEO requires real audit depth.

But schema markup? You can add dozens of structured data blocks in an afternoon. You can generate a report showing every schema type implemented. You can create a before-and-after deliverable that looks impressive in a slide deck. It feels technical. It feels like AI-adjacent work. And most clients have no framework to evaluate whether it actually helped anything.

So the narrative was born: “Schema markup helps AI systems understand your business.” It spread fast because it sounded plausible to people who did not know better, and because it gave agencies something concrete to charge for.

“When an industry lacks real products, it manufactures perceived complexity. The ‘schema for AI’ pitch is a perfect example of selling process as outcome.”

The problem is that it is not true. And now there is research to confirm it.

What the Research Actually Shows: Schema Does Not Improve AI Citations

Research from Ahrefs has confirmed that implementing schema markup does not meaningfully improve a website’s chances of being cited, mentioned, or extracted by AI systems like Google AI Overviews, ChatGPT, Gemini, or Perplexity. AI citation is driven primarily by content quality, topical authority, third-party signals, and entity recognition, not structured data markup.

Ahrefs conducted analysis specifically looking at the relationship between structured data and AI Overview appearances. The findings were consistent with what many of us had already suspected from practical experience: schema markup showed no meaningful correlation with AI citation rates. Sites without any schema markup were appearing in AI Overviews just as readily as sites with extensive structured data implementations, when their content quality and authority were comparable.

This makes complete sense once you understand how large language models actually work.

LLMs, including the retrieval systems feeding tools like Google’s AI Overviews, Perplexity, ChatGPT Browse, and Bing Copilot, are not parsing your JSON-LD blocks to figure out what your business does. They are reading your actual content. They are evaluating signals from across the web, mentions in third-party sources, backlink profiles, brand recognition, content coherence, and contextual authority. A well-written, clearly structured paragraph that directly answers a user’s question is infinitely more valuable for AI retrieval than a perfectly formatted Organization schema block.

Why LLMs Do Not Need Schema to Understand Your Website

This is the part that the “AI optimization” agencies deliberately obscure, because if clients understood it, the pitch would fall apart immediately.

Large language models are extraordinarily good at reading natural language. That is literally what they were trained on, billions of pages of plain text. When a well-written webpage clearly explains what a business does, where it operates, what problems it solves, and who it serves, an LLM understands all of that without a single line of structured data.

Think about how Google AI Overviews, ChatGPT, or Perplexity actually select sources to cite:

• Is the content authoritative on this specific topic?
• Is the information corroborated by signals from other credible sources?
• Does the content directly and clearly answer the question being asked?
• Does the entity behind the content have recognized authority in this domain?
• Are there third-party mentions, citations, and links pointing to this content?

None of these criteria involve schema markup. Not one.

An LLM does not look at your FAQPage schema and decide you are more trustworthy. It looks at your actual content, how well it is written, how thoroughly it covers a topic, whether other credible sources reference your work, and whether your entity has a recognizable footprint across the web. That is E-E-A-T applied at the machine-learning layer, not structured data.

“If your content is unclear, schema will not save you. If your content is excellent, schema is irrelevant to whether AI systems find and cite you. The substance is everything. The markup is decoration.”

The Meta Keywords Parallel: A History Lesson the Industry Refuses to Learn

For anyone who was doing SEO in the late 1990s and early 2000s, the meta keywords tag story is painfully familiar. Meta keywords was a legitimate HTML element designed to help early search engines understand the topical focus of a webpage. It worked, briefly, until webmasters started stuffing it with hundreds of keywords, competitor brand names, and completely irrelevant terms to game rankings.

Search engines responded the only way they could. Google officially confirmed it had stopped using the meta keywords tag as a ranking signal. Other major engines followed. Today, the meta keywords tag is a relic, completely ignored by every search engine that matters, existing only as dead HTML that some CMS templates still generate out of habit.

Schema markup is following an uncomfortably similar trajectory.

The abuse is already well underway. Sites are adding schema types that have no relationship to their actual content. Businesses are implementing FAQPage schema for questions that are not genuinely answered on the page. Review schema is being added to pages that contain no actual reviews. Article schema is being applied to service pages. The entire structured data ecosystem is becoming polluted with noise generated by agencies trying to show deliverables rather than actual optimization.

Google has already started restricting which schema types trigger rich results. The FAQ rich result was significantly curtailed, appearing only for authoritative government and health sites in most cases now. The How-To rich snippet was pulled from desktop results. These are not coincidences. These are responses to abuse.

If the current trend of mass schema implementation continues without purpose, the logical endpoint is what happened to meta keywords: search engines and AI systems simply begin ignoring structured data entirely, or weight it so minimally it becomes meaningless.

Where Schema Markup Is Actually Legitimate and Useful

I want to be fair here, because writing off schema entirely would itself be bad advice. There are specific, well-defined use cases where schema implementation genuinely matters and produces measurable results.

Legitimate Schema Use Cases

The pattern is obvious. Schema helps with display enhancements in traditional search results. It is a communication layer between your site and Google’s rendering engine, not an intelligence layer between your content and AI models.

Use schema where it makes contextual sense. Implement it correctly and only where the content on the page actually supports it. Do not implement it because someone told you it would make ChatGPT mention your brand more often. That is not how any of this works.

What Actually Improves AI Citation and Visibility

AI systems cite sources based on content quality, topical authority, third-party mentions, backlink credibility, and how clearly and directly the content answers user questions. Improving AI visibility requires building genuine authority through expert-authored content, earning mentions from credible third-party sources, and maintaining a strong entity presence across the web.

If your goal is to appear in AI Overviews, be cited by ChatGPT, or get mentioned in Perplexity answers, the actual playbook looks like this:

1. Write content that directly and completely answers specific questions. AI retrieval systems favor content that provides clear, quotable answers without burying the point in filler.
2. Build topical authority through content depth. Sites that comprehensively cover a subject from multiple angles are more likely to be recognized as authoritative sources by LLMs.
3. Earn third-party mentions and citations. When other credible websites reference your content, link to your pages, or mention your brand, AI systems receive corroborating signals that strengthen your authority.
4. Strengthen your entity footprint. Having a consistent, verifiable presence across Google Business Profile, Wikipedia references, industry directories, social platforms, and credible publications creates a recognizable entity that AI systems can confidently associate with a topic.
5. Use clear, structured writing. Proper HTML heading hierarchy, logically organized content, and direct language make it easier for AI systems to extract and summarize your information, no schema required.

Notice what is not on that list. Schema markup. Because it does not belong there.

The Broader Problem: AI-Washing in the SEO Industry

The schema-for-AI pitch is a symptom of a larger issue I call AI-washing in the SEO industry. As AI tools became the dominant conversation in tech, a segment of the SEO market rushed to rebrand existing services under an AI optimization umbrella, often with zero additional value delivered to clients.

This is how it typically plays out:

• An agency rebrands their standard SEO package as “AI SEO” or “GEO” (Generative Engine Optimization)
• They add schema implementation as a core deliverable
• They produce reports showing schema coverage across the site
• They cite vague improvements in “AI readiness” without defining what that means or how it is measured
• Clients pay premium rates for what amounts to a technical decoration project

The business model works because most clients lack the technical background to evaluate the claims. “We optimized your schema for AI retrieval” sounds sophisticated. The fact that it is largely meaningless for the stated purpose is not something the deliverable report will ever mention.

“Selling schema implementation as AI optimization is the SEO equivalent of selling someone a fresh coat of paint on a car and telling them it improves fuel efficiency. The paint is real. The fuel efficiency claim is not.”

Myths vs. Facts: Clearing Up the Schema Markup Confusion

Myth: Schema markup helps AI systems understand your business better

Fact: LLMs understand your business from your written content, third-party mentions, and entity signals. They do not rely on JSON-LD blocks to determine what your company does or whether your content is authoritative.

Myth: Adding more schema types increases your AI citation chances

Fact: Ahrefs research found no meaningful correlation between schema implementation and AI Overview inclusion rates. More schema does not equal more AI visibility.

Myth: Schema markup is a ranking factor

Fact: Google has repeatedly confirmed that schema markup is not a direct ranking signal for standard organic results. It influences rich result eligibility, not ranking position.

Myth: Implementing schema is low-risk and always beneficial

Fact: Implementing irrelevant or inaccurate schema can trigger Google manual actions for structured data spam. It also contributes to the ecosystem degradation that may lead Google to reduce schema’s influence entirely.

Myth: Sites with comprehensive schema coverage outperform sites without it

Fact: Sites perform well because of content quality, authority, and technical fundamentals. Schema enhances how that performance is displayed, not whether it exists.

My Practical Recommendation on Schema Markup SEO

Here is where I land after years of working in this industry and watching these trends cycle through:

Implement schema markup where it is contextually appropriate and technically correct. If you run a local service business, use LocalBusiness schema with accurate NAP data. If you have genuine product reviews, implement Product and AggregateRating schema. If you publish recipes, use Recipe schema. These implementations have clear, demonstrable use cases tied to rich result eligibility.

Do not implement schema as an AI optimization strategy. Do not pay an agency premium rates for schema-heavy deliverables sold under an AI optimization banner. And if an agency is telling you that adding structured data will improve your ChatGPT mentions or Google AI Overview appearances, ask them to show you the evidence. Not case studies they wrote themselves. Actual third-party research. They will not be able to produce it, because it does not exist.

Spend that budget instead on content that genuinely answers questions your audience is asking. Build relationships that generate real backlinks and brand mentions. Create an entity presence that AI systems can recognize and trust. These are the signals that move the needle on both traditional SEO and AI retrieval, and they always will be, regardless of what the next shiny tactic turns out to be.

Is Schema Markup Going to Become the Next Meta Keywords Tag?

Possibly. The conditions are eerily similar.

Meta keywords died because widespread abuse destroyed its signal value. Google had two choices: attempt to clean up the signal or ignore it entirely. Ignoring it was simpler and more reliable, so that is what happened.

Schema markup is facing the same dynamic. The signal is becoming noisier as more sites implement it incorrectly, irrelevantly, or deceptively. Google has already begun reducing the visibility of rich results for specific schema types as a response. If the trend continues, we may reach a point where Google’s systems simply down-weight structured data across the board, not because schema itself is bad, but because the ecosystem of implementation has become too polluted to trust.

That would be a genuine loss for the legitimate use cases. But it would be a predictable outcome of an industry that never learned its lesson from 2003.

Working With an SEO Expert Who Tells You the Truth

One of the rarest things in this industry is finding someone willing to tell you what does not work, not just what they can sell you. My approach at Affordable SEO Expert has always been to focus on strategies with actual, measurable ROI, content authority, technical fundamentals, link equity, and entity building, rather than chasing tactics that look impressive in reports but produce nothing in reality.

If you have been sold an AI optimization package heavy on schema markup deliverables and light on real results, I am happy to take an honest look at what your site actually needs and where your budget would produce genuine returns. Good SEO is not complicated. It is just rare to find people willing to do it without wrapping it in unnecessary complexity.

Frequently Asked Questions About Schema Markup and AI Optimization

Does schema markup help with Google AI Overviews?

No, schema markup does not meaningfully improve your chances of appearing in Google AI Overviews. Research from Ahrefs confirmed no significant correlation between structured data implementation and AI Overview inclusion. Google’s AI systems select sources based on content quality, topical authority, and third-party credibility signals, not schema presence.

Is schema markup a Google ranking factor?

Schema markup is not a direct ranking factor for standard organic search results. Google has confirmed this multiple times. Schema’s primary function is to make content eligible for rich results, such as star ratings, FAQ dropdowns, and event listings. It influences how your result is displayed, not where it ranks.

Why do so many SEO agencies sell schema markup as an AI optimization service?

Schema markup is easy to implement, visually impressive in reports, and sounds technically sophisticated to clients unfamiliar with how AI systems work. It provides a tangible deliverable that agencies can show as proof of work, even though the correlation between schema implementation and AI citation improvement is not supported by independent research.

Can too much schema markup hurt your website?

Yes. Implementing inaccurate, misleading, or irrelevant schema markup can trigger a Google manual action for structured data spam. Additionally, widespread schema abuse is likely contributing to Google’s gradual restriction of rich result types, meaning over-implementation at an industry level is actively degrading the value of structured data for everyone.

What actually improves AI citation and AI search visibility?

AI systems prioritize content that directly and clearly answers user questions, has topical depth and authority, and is corroborated by third-party signals like backlinks, brand mentions, and entity recognition. Writing expert-level content, earning credible inbound links, and building a consistent entity presence across the web are the most effective strategies for improving visibility in AI-generated answers and citations.