There is a pattern in SEO that repeats itself every few years. A technical feature gets introduced for a legitimate, narrow purpose. Then a wave of consultants, agencies, and self-proclaimed optimization experts start treating it like a ranking superpower. Clients get sold on it. Everyone adds it everywhere. Then the signal either gets devalued, abused into irrelevance, or quietly deprecated. We watched it happen with meta keywords. We are watching it happen right now with schema markup.

What makes this cycle especially frustrating today is that it is being accelerated by an entirely new category of snake oil: the fake “AI optimization” industry. These companies needed something tangible to sell. They needed something technical enough to look sophisticated but simple enough to actually deliver. Schema markup became their answer. And now the entire conversation around structured data has been polluted by bad-faith marketing and genuinely misleading claims about what schema actually does.

We want to set the record straight on schema markup SEO, what structured data is actually for, why it does not help with AI citations the way people claim, and why this particular trend might be heading toward the same graveyard as meta keywords.

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

Schema markup is a form of structured data vocabulary, developed through Schema.org, that helps search engines understand specific entities and content types on a page. Its primary, well-documented benefit is enabling rich results in Google Search, such as star ratings, FAQ dropdowns, recipe cards, and event listings. It was never designed as a general ranking signal or an AI optimization tool.

Schema markup is a set of standardized labels you add to your HTML to help search engines identify the type of content on a page. It uses a shared vocabulary maintained at Schema.org, a collaborative effort originally launched by Google, Bing, Yahoo, and Yandex. When you implement a Product schema, you are telling Google explicitly that this page describes a product, here is its price, and here is its rating. When you implement LocalBusiness schema, you are labeling your name, address, and phone number in a machine-readable way.

The core use case has always been rich snippets. That is it. Google uses structured data to generate enhanced visual results in the SERP, things like star ratings on review pages, breadcrumb trails, how-to steps, FAQ accordions, and event dates. These are genuinely useful. They improve click-through rates on the right pages. They serve a real purpose when implemented correctly and in the right context.

What schema was never designed to do is improve organic rankings broadly, replace good content strategy, or signal authority to large language models crawling the web.

The Fake AI Optimization Industry Needed a Product to Sell

When the hype around AI-driven search exploded, a new category of agencies and freelancers emerged almost overnight. They called themselves “AI optimization specialists,” “GEO consultants,” “AEO experts,” and a dozen other variations of the same invented credential. The problem they faced immediately was obvious: what exactly do they deliver?

Genuine AI visibility work is hard to package. It involves content quality, topical authority, brand presence, third-party mentions, genuine expertise signals, and the trust that builds over the years. None of that fits neatly into a deliverable you can send a client at the end of the month.

Schema markup fits perfectly. It is technical enough that most clients do not fully understand it. It can be added to a website in bulk relatively quickly. It produces a report of implementations that looks like work. And it has just enough of a legitimate SEO reputation that you can invoke it without outright lying, even if you are dramatically overstating what it accomplishes.

“Schema markup became the perfect fake deliverable for the AI optimization grift. It looks technical, it can be bulk-implemented fast, and most clients will never know it did nothing.”

So the pitch became: add schema markup everywhere across your site and you will be better positioned for AI search, AI citations, Google AI Overviews, and generative answers. This framing spread across LinkedIn posts, YouTube videos, and agency service pages at remarkable speed. And because it had the veneer of technical legitimacy, a lot of well-meaning businesses bought into it.

The Ahrefs Study and What the Data Actually Shows

Research from Ahrefs found no meaningful correlation between schema markup implementation and increased AI citations or mentions in AI Overviews. The study examined a large sample of URLs and found that structured data presence did not predict whether a page would be cited in generative AI responses, confirming that LLMs do not rely on schema to understand or surface content.

Ahrefs has been among the most rigorous voices in applying actual data analysis to AI search claims, and their research into schema markup and AI citations is one of the clearest takedowns of the AI optimization grift available. The findings were straightforward: pages with rich structured data were not more likely to be cited in AI-generated answers than pages without it. The correlation simply was not there.

This should not be surprising to anyone who understands how large language models actually work. LLMs do not parse JSON-LD blocks in the way a search engine rich result system does. They are not scanning your FAQPage schema and deciding your content is more trustworthy because you labeled your questions correctly. They read language. They understand context. They have been trained on massive corpora of text and have internalized patterns of meaning, entity relationships, and factual associations in ways that make the presence or absence of a schema tag largely irrelevant to whether they surface your content.

The factors that actually drive AI citations are well-established and have nothing to do with structured data:

• High-quality, clearly written content that directly answers questions
• Strong third-party signals including mentions, backlinks, and references from authoritative sources
• Brand entity recognition across multiple platforms and data sources
• Consistent business information across the web, not just on-site schema
• Topical authority built through depth and breadth of coverage
• Content that is already indexed and ranked well in traditional search

None of those are things you can shortcut by adding JSON-LD to your page templates.

LLMs Don’t Need Schema to Understand Your Business

Large language models understand content through natural language processing, not structured data tags. An LLM can determine what a business does, what it sells, where it is located, and who its customers are simply by reading well-written page content. Schema markup adds zero interpretive value for LLMs because they are not designed to process it as a semantic signal the way traditional search engine crawlers use it for rich results.

One of the fundamental misunderstandings driving the schema-for-AI narrative is a confusion between how traditional search crawlers work and how language models work. Google’s rich result system processes structured data in a very specific, programmatic way. It looks for defined schema types, extracts specific property values, and uses that structured information to generate enhanced SERP features. That system genuinely needs schema markup to work correctly.

Language models operate completely differently. They were trained by reading text, enormous volumes of it, and learning statistical and semantic relationships between words, sentences, entities, and concepts. When a language model encounters your business’s website, it does not run a schema validator. It reads your content. If your content clearly explains that you are a plumbing company in Chicago that specializes in emergency pipe repairs, the model understands that, because the sentence says it clearly.

Adding LocalBusiness schema to that same page does not make the model understand it better. The content already communicated everything the model needs. What actually influences whether that model surfaces your business in a response is whether it encountered your name, services, and location across multiple credible sources during training and indexing. That means Google Business Profile, industry directories, local news mentions, review platforms, cited articles, and social media presence.

“An LLM doesn’t need you to label your content with JSON-LD. It needs you to be worth talking about. Those are fundamentally different problems.”

The companies selling schema as an AI optimization service are, whether deliberately or through ignorance, conflating two entirely different systems and charging clients for work that addresses neither.

Schema Markup Overuse: Exactly How the Abuse Is Happening

When we say schema is being abused, we are not talking about edge cases. We are seeing patterns across industries that are directly parallel to the meta keywords abuse that caused Google to deprecate that signal entirely.

What Legitimate Schema Use Looks Like

Implemented correctly, schema markup is a precise, targeted tool. You add Product schema to product pages where you want price and availability shown in search results. You add Recipe schema to recipe pages. You add LocalBusiness schema to your homepage or contact page to reinforce your NAP data. You implement HowTo schema when you have a genuine step-by-step instructional page. The implementation matches the actual content type and the goal is a specific rich result.

What Abuse Looks Like Right Now

• Adding FAQ schema to every single page on a site regardless of whether the content is actually structured as FAQs
• Implementing Article schema with inflated or irrelevant authorship claims
• Adding Organization or LocalBusiness markup on dozens of pages where it serves no functional purpose
• Stacking multiple schema types on pages where they do not apply, to create the appearance of rich structured data coverage
• Adding SpeakableSpecification schema across pages not optimized for voice search, under the guise of AI readability
• Implementing schema for entity types the business does not actually represent
• Marking up subjective claims as objective structured data

The common thread is volume over relevance. Agencies are bulk-implementing schema across sites to generate deliverable reports rather than to serve any legitimate technical or user experience purpose. And clients, unless they understand the mechanics, have no reason to question it.

Schema Markup vs. Meta Keywords: A Parallel Worth Taking Seriously

The parallel is not rhetorical. Google has already started walking back support for certain schema types. FAQ rich results were significantly restricted, no longer showing for most websites. HowTo rich results were similarly curtailed for desktop results. These are not coincidental. Google reduces or removes rich result support precisely when the signal loses reliability due to overuse and misuse. That trajectory is continuing.

“Schema markup is not the new SEO superpower. It is the new meta keywords tag, and we are watching the exact same cycle play out in real time.”

What Google Has Actually Said About Schema and Rankings

Google has been remarkably consistent and remarkably ignored on this point. John Mueller and other Google Search representatives have stated clearly and repeatedly that structured data is not a ranking signal. It does not cause pages to rank higher in organic results. It enables rich results when the content qualifies and the implementation is correct. That is the entirety of its documented benefit in traditional search.

Google’s own documentation explicitly states that structured data “can make it easier for Google to understand what your page is about” in the context of specific features, not general crawlability or ranking. For any well-written page, Google already understands what it is about. The schema implementation is confirmation for specific SERP features, not comprehension assistance.

Yet the myth persists because it is useful to those selling it. A client hears “helps Google understand your content better” and assumes that means higher rankings. That assumption is wrong, but it is a profitable misunderstanding for the agency collecting the retainer.

The Third-Party Signal Reality That Gets Ignored

Here is what actually influences how AI systems understand and represent your business: the entire external web of signals that references you.

Google’s AI Overviews, ChatGPT browsing, Perplexity, Gemini, and every other AI-driven search or answer system is drawing from a vast pool of information that extends far beyond your website. They are looking at:

• Your Google Business Profile and the data consistency across local citations
• Reviews on Google, Yelp, Trustpilot, and industry-specific platforms
• Articles, press mentions, and blog posts that reference your brand
• Backlinks from authoritative domains that establish topical relevance
• Wikipedia presence or mentions where applicable
• Social media profiles and the entity coherence they establish
• Your content’s ability to rank organically, since AI systems disproportionately surface already-ranking pages
• Brand mentions that appear without links, which are still processed as co-occurrence signals

Notice what is not on that list. Schema markup. Because a JSON-LD block on your website, however perfectly formatted, does not add a single data point to the external web of signals that AI systems rely on to understand your brand’s authority, reach, and relevance.

If an agency is selling you schema implementation as an AI optimization strategy without spending significant time on your content quality, link profile, brand mentions, and digital presence, they are solving a problem that does not exist while ignoring the ones that do.

When Schema Markup Does Make Sense (The Legitimate Cases)

We are not arguing that schema markup has no value. We are arguing that its value is specific, limited, and being dramatically misrepresented. There are genuine scenarios where implementing structured data produces real, measurable benefits.

Cases Where Schema Provides Real Value

• E-commerce product pages: Product schema with price, availability, and aggregate ratings can produce rich results that meaningfully improve click-through rates for competitive product searches.
• Recipe websites: Recipe schema is one of the most visually rich implementations in Google Search and directly drives traffic for culinary content publishers.
• Event listings: Event schema helps concert venues, conference organizers, and local businesses surface date and ticket information directly in search results.
• Local business pages: A single, accurate LocalBusiness implementation on your primary contact or homepage page reinforces your NAP data in a machine-readable format. One implementation, done correctly, is sufficient.
• Review and rating aggregators: Aggregate rating schema for genuine review content provides click-through rate benefits in competitive categories.

Cases Where Schema Adds Nothing

• Service pages that do not have a qualifying schema type
• Blog posts that do not qualify for specific article-type rich results
• About pages, team pages, or legal pages with no rich result use case
• Any page where the goal is “AI optimization” rather than a specific SERP feature

The test for whether to add schema is simple: is there a specific rich result this implementation could generate, and would that rich result improve the user experience for this particular page? If the answer is yes, implement it carefully. If the answer is no, leave it out.

Why This Matters Beyond Just Wasted Budget

The consequences of the schema oversell go further than clients paying for unnecessary implementations. There are real harms happening across the ecosystem.

First, businesses are being given a false sense of security. They believe their AI visibility is being addressed when it is not. The actual work that would improve their presence in AI-generated answers, which means building genuine authority, earning third-party coverage, improving content quality, and expanding their digital footprint, is going undone while agencies are collecting fees for structured data audits.

Second, the signal itself is being degraded. The more widespread and indiscriminate schema implementation becomes, the less reliable it is as a source of information for Google’s systems. This has already played out with FAQ and HowTo rich results, and it will likely continue with other schema types. Legitimate businesses implementing schema correctly for genuine rich result purposes will eventually find that signal weakened by the volume of noise around it.

Third, it contributes to a broader industry credibility problem. When the “AI SEO” space is dominated by claims that have no empirical basis, it makes it harder for businesses to identify legitimate expertise. The signal-to-noise ratio in SEO has always been challenging. Manufactured hype around schema for AI optimization makes it worse.

How to Actually Evaluate Any Schema Markup Recommendation

If someone recommends adding or expanding schema markup on your site, here are the questions that separate legitimate technical advice from filler deliverables:

1. What specific rich result is this implementation targeting? There should be a direct answer naming a specific SERP feature, not a vague reference to “better understanding” or “AI readiness.”
2. Does the content on this page actually qualify for that rich result? Google has eligibility requirements for most rich results. The content needs to match the schema type genuinely.
3. Is there a Google Search Console report or testing tool result that validates this implementation? Google provides explicit tools for validating structured data. Any agency implementing schema should be using them.
4. What is the expected change in CTR or impression share from this implementation? If there is no answer, there is no strategy.
5. How does this relate to AI citation improvement? If the answer invokes schema as an AI optimization tool without citing specific, credible research, treat that as a red flag.

Our Take: Where Schema Is Headed

The trajectory of schema markup, given the current pattern of abuse, is not encouraging for those who overinvest in it based on inflated claims. Google has clearly demonstrated that it will reduce or remove rich results support for schema types that are being gamed. It has already done this twice with FAQ and HowTo results. Additional restrictions are not speculative; they are the logical continuation of a pattern Google has already established.

The meta keywords parallel is not perfect. Schema markup has more legitimate uses than meta keywords ever did. But the social dynamic is identical: a signal with a narrow, specific use case gets inflated into a general-purpose optimization strategy by people who profit from that inflation, and the signal deteriorates in value as a result. We have seen this movie before. We know how it ends.

At Marketing 1on1, our approach to schema has always been exactly what the evidence supports: implement it precisely where it produces rich results, ignore it everywhere else, and never, under any circumstances, sell it as an AI optimization strategy to clients who deserve honest advice.

Stop Letting the Hype Machine Define Your SEO Strategy

Schema markup is a useful, specific tool for achieving rich results in Google Search. It is not a ranking signal. It is not an AI citation driver. It is not a shortcut to appearing in Google AI Overviews or getting cited by ChatGPT. The research says so. Google’s own documentation says so. And anyone with genuine technical SEO experience working in the field knows this.

The companies telling you otherwise needed a product to sell when the AI optimization goldrush started. Schema markup was convenient, looked technical, and could be bulk-delivered quickly. That does not make it effective for the purposes being claimed. It makes it a perfect vehicle for billing hours on work that does not move the needle.

Real AI visibility improvement is slower, harder, and more expensive in terms of genuine effort. It means creating content that is genuinely worth citing. It means building a brand presence across the web that AI systems encounter repeatedly and consistently. It means earning mentions in publications, directories, and platforms that carry real authority. None of that shows up in a tidy monthly deliverable. All of it actually works.

If your current SEO partner is leading with schema as an AI optimization strategy, ask them to show you the data. Ask them which specific studies support the claim. Ask them which rich results your schema implementations have generated and what the CTR impact was. If the answers are vague, you already have your answer.

We work with businesses that want honest, evidence-based SEO strategy, not deliverables designed to look like progress. If that sounds like what you are looking for, contact us today, and let’s talk about what actually moves the needle.

Frequently Asked Questions About Schema Markup

Does schema markup improve Google rankings?

No. Google has confirmed multiple times through official documentation and spokesperson statements that schema markup is not a ranking signal. Structured data does not cause pages to rank higher in organic search results. Its documented benefit is enabling rich results, which are enhanced visual features in the SERP that can improve click-through rates for qualifying pages. Any claim that schema improves rankings directly is not supported by evidence.

Does schema markup help with AI citations in Google AI Overviews or ChatGPT?

No meaningful correlation between schema markup and AI citations has been established by credible research. Studies including analysis from Ahrefs found that structured data presence does not predict whether a page will be cited in AI-generated answers. LLMs understand content through natural language processing, not by parsing schema tags, and AI citation is driven by content quality, third-party authority signals, and brand presence across the web.

What is schema markup actually used for?

Schema markup is structured data vocabulary from Schema.org used to enable rich results in Google Search. Legitimate applications include product schema for e-commerce rich results showing price and ratings, recipe schema for recipe cards, event schema for date and ticket information, and LocalBusiness schema to reinforce business information in search. Each implementation should target a specific, eligible rich result feature. Schema outside of these use cases provides no documented benefit.

Why is schema markup being compared to the meta keywords tag?

The comparison reflects a recurring SEO pattern where a signal with a legitimate, narrow use case gets inflated into a general optimization strategy, abused at scale, and eventually devalued. Meta keywords were deprecated by Google after widespread stuffing abuse. Schema markup is currently being over-implemented across sites as a fake AI optimization tactic, and Google has already reduced rich result support for FAQ and HowTo schema types in response to abuse. The same degradation pattern is underway.

How many schema types should a typical business website have?

The number should be determined entirely by how many pages have genuine, eligible rich result use cases. For most small to mid-size businesses, a single accurate LocalBusiness implementation on the homepage or contact page, combined with any product or service-specific schema on qualifying pages, covers the legitimate scope. There is no SEO or AI benefit to adding schema across all pages as a blanket practice. Precision is more valuable than volume.