Schema markup is one of the most consistently recommended tactics in technical SEO, and for good reason. It has a 10+ year track record of helping Google surface rich results, trigger featured snippets, and improve SERP feature eligibility, benefits that remain fully intact today. As AI Search platforms have grown in prominence, a natural question has followed: if schema helps traditional search engines understand content, does it also help AI Search platforms?
To find out, OtterlyAI ran a controlled experiment to test that assumption. We implemented five types of schema markup on live pages, directly prompted on 7 major AI Search platforms to retrieve that structured data, and tracked brand coverage across 319 prompts over a 3 month observation window.
One important caveat before going further: everything in this article is based on data from the OtterlyAI website specifically. OtterlyAI is a SaaS brand, not an ecommerce business. These findings may not translate directly to every industry or site type, and we’d encourage you to treat them as a starting point for your own testing rather than a universal rule.
Key Findings (TL;DR)
6 out of 7 AI Search platforms were unable to fetch or correctly interpret schema markup when directly asked. Only Gemini retrieved the correct JSON-LD data.
Google AI Mode hallucinated schema types that were not implemented, including a “Service” schema that did not exist on the page.
Increase detected on Google AI Mode and AI Overviews: after 3 months, SERP features increased by +377% and AI Overviews by +1,500%. Schema still delivers meaningful SEO results.
Impact on other AI Search platforms: For AI Search platforms beyond Google’s own ecosystem, however, its direct impact appears to be much more limited
AI Search Engines don’t need schema markup the way traditional crawlers do. Schema was invented for algorithms that couldn’t handle unstructured text. AI Search is built to do exactly that.
UseAI Merchant programs for products. ChatGPT’s Merchant Program & Perplexity’s Buy with Pro are better alternatives than schema for AI platforms. The future is AI buying products directly.
Why Studying Schema Markup Matters for GEO
AI Search visibility is increasingly determined by citation selection rather than page ranking. When ChatGPT, Perplexity, or Google AI Overviews answer a query, they retrieve and cite sources ,and the logic that determines which sources they pull from is fundamentally different from the logic that governs traditional SEO.
Schema markup sits at the intersection of both worlds, which makes it worth testing carefully. It was designed to annotate structured meaning on top of raw content ,to help parsers that couldn’t read unstructured text understand what a page was about. If AI platforms are now doing that parsing natively, as large language models are designed to do, then schema’s role in GEO may be very different from what it is in traditional SEO.
That’s the question this experiment was designed to answer with real data.
Scope of Study
Everything in this article is based on OtterlyAI’s own website and brand data. This is not an industry-wide study, it is a controlled experiment run on a single SaaS brand’s web presence, across a defined set of pages and a specific 3 month window. The findings are most applicable to SaaS, content-focused, and service-category brands. If you run an ecommerce site where product, pricing, and review schemas have well-established behavior, your results may differ, and we’d strongly encourage running your own test before drawing conclusions.
This GEO Experiment was conducted from 7th of December 2025 to 7th of March 2026 across seven AI platforms: ChatGPT, Google AI Overviews, Google AI Mode, Perplexity, Microsoft Copilot, Gemini, and Claude.
The following Schema Markup types were implemented and tested:
FAQ Schema
Organization Schema
SoftwareApplication Schema
Review Snippets Schema
Article / BlogPosting Schema
On Dec 7th. Schema Markup was placed on the entire Otterly.ai website. For the purpose of this GEO experiment, we’ll limit the scope to discussing the impact on the following URLs:
A direct capability test: Each AI platform was prompted to fetch and return the raw schema markup code from both URLs, testing whether AI systems can access and interpret structured data at all.
Impact on AI brand visibility test: OtterlyAI’s Brand Coverage was tracked across 319 prompts in the US market using OtterlyAI’s AI Search Monitoring, from November 1, 2025 through March 2026, with schema implemented on December 7, 2025 as the intervention point. Competitor brand coverage was tracked simultaneously to control for algorithmic shifts.
Unique information test: A piece of information was added to a specific page via an FAQ Schema that could not exist anywhere else on the web.
One interpretive note applies throughout: OtterlyAI is a SaaS brand, not an ecommerce business. Schema behavior may differ in ecommerce contexts where product, pricing, and review schemas have more established citation patterns. These findings are strongest for SaaS, content, and service-category brands.
Methodology
Step 1: Select Schema Types That Represent Real GEO Use Cases
Three schema types were selected to cover the most common structured data patterns relevant to a SaaS and content brand:
FAQ Schema: Annotates question-and-answer content ,directly relevant to how AI platforms decompose informational queries
Article / BlogPosting Schema: Provides authorship, publication date, and content-type signals for editorial content
These types were chosen because they map directly to the kinds of pages most brands invest in for GEO ,not because they were expected to succeed.
Step 2: Implement All Schema in JSON-LD
All schema was implemented using JSON-LD format ,the approach recommended by schema.org and Google’s developer guidelines, placed inside <script type=”application/ld+json”> tags in the <head> of both target pages. Implementation was completed on December 7, 2025.
https://otterly.ai/agencies
Pro Tip: Use tools like schema validator to check if your structured data is working properly
Step 3: Configure Brand Monitoring Baseline
An OtterlyAI Brand Report was set up prior to the experiment tracking 319 prompts in the United States market, covering all major AI engines. The November 2025 period established pre-implementation brand coverage baselines for OtterlyAI and all detected competitors.
Step 4: Prompt Each AI Platform to Fetch the Schema
Each major AI Search platform was individually asked:
“Fetch me the Schema Markup code from this page: [URL]”
Both target URLs were tested across all seven platforms. The goal was to answer a foundational question before measuring downstream effects: can AI platforms actually read schema markup? If they cannot access the raw JSON-LD, the argument that schema influences AI citation behavior collapses at the architectural level.
Step 5: Measure Brand Coverage Change and Control for Algorithmic Shifts
Any change in OtterlyAI’s brand coverage after December 7 was compared against competitor trajectories over the same period. A simultaneous lift across multiple brands ,none of whom had implemented schema changes ,would indicate an algorithmic shift rather than a schema effect. This control is essential to avoid misattributing platform-level changes to a specific technical intervention.
GEO Experiments & Results
Test 1: Ask Directly if AI Platforms Can Actually Read Schema Markup?
Results: Six out of seven platforms failed.
The direct fetch test returned a decisive result.
AI Search Platform
Could it fetch Schema Markup when asked?
Gemini
✅ Yes ,returned the correct JSON-LD schemas accurately
Claude
❌ No, claimed no schema markup was present on the page
ChatGPT
❌ No, could not detect schema in the accessible source
Perplexity
❌ No ,server blocked the fetch; redirected to manual extraction instructions
Google AI Mode
⚠️ Partial, returned schema types but hallucinated incorrect schemas not present on the page
Google AI Overviews
⚠️ Google AI Overviews didn’t return a fetch result. However, schema markup appears to influence how often AI Overviews are triggered for a query, one of the stronger indirect arguments for keeping structured data on your pages.
Microsoft Copilot
❌ No ,could only access readable text, not raw HTML
Claude
Claude was not able to fetch the Schema Markup code and even claimed there was no Schema Markup present on both attempts
ChatGPT
ChatGPT had similar results, it also couldn’t fetch the schema markup data when requested.
I did one more attempt using ChatGPT’s deep research mode.
Perplexity
Perplexity showed very similar results. Also did not manage to fetch the requested Schema Markup fetching.
Google AI Mode
Google AI Mode did mention multiple schemas but also included the “service schema” which was not actually on the page.
Google AI Overviews
Since Google AI Overviews is a native part of the classic Google Search Engine. It does not fetch an answer in a similar way that the other AI Search platforms do. It did not return an answer.
Microsoft Copilot
Microsoft Copilot was also not able to fetch the requested schema markup
Gemini
In both instances, Gemini 3 was able to fetch the schema markup when requested.
The single platform that succeeded, Gemini, accurately returned the BlogPosting and WebPage schemas for the blog post URL, including headline, description, author, publisher, keywords, and date fields. Every other platform either failed silently, redirected to manual instructions, or produced something worse than failure.
The Google AI Mode Exception Is Worth Examining
Google AI Mode didn’t just fail, it hallucinated. It returned a description of plausible-sounding schema types and described them confidently in natural language, but included a “Service” schema that was never implemented on the page.
A platform that confidently returns wrong structured data is not reading schema. It is generating text that sounds like schema. That distinction matters significantly for anyone making GEO decisions based on the assumption that schema is being consumed.
Why Many AI Platforms Won’t Process Schema Markup
Many of the HTML-to-Markdown and extraction pipelines used in AI Search workflows either ignore or destroy <script>tags entirely or treat them as low‑value content, which means JSON-LD schema inside <script type=”application/ld+json”>is often removed before the model ever sees the page.
AI Search Platforms are built to process unstructured text. Schema markup was invented for parsers that couldn’t.
Schema was created in 2011 by Google, Bing, and Yahoo as a solution to a specific problem: search engine crawlers at the time couldn’t reliably interpret the semantic meaning of unstructured web content. Structured annotations gave those parsers explicit labels, “this is a product,” “this is a review,” “this is a FAQ”, that they couldn’t reliably infer from prose alone.
A language model doesn’t need a FAQPage annotation to understand that a page contains questions and answers. It can read the content directly and infer the structure. Schema was a workaround for algorithmic limitations that AI Search Platforms don’t have.
Test 2: Did Schema Implementation Increase AI visibility?
Schema markup was implemented on December 7, 2025. OtterlyAI’s brand coverage across 319 tracked prompts was monitored before and after that date. Below is a breakdown of AI visibility split by AI Search Platforms. To reduce external signals the scope was reduced to 3 studied pages where Schema Markup was implemented.
Results
Google AI Mode and AI Overviews were the only 2 platforms that saw an increase in AI visibility for the tested pages.
The AI Overviews increase, in particular, was substantial, consistent with schema markup contributing to SERP feature eligibility and the indexing signals that feed into whether AI Overviews are triggered at all.
All other platforms saw a little impact in AI visibility for the tested pages.
AI Search Platform
Citation Increase or Decrease?
% change (3 months)
ChatGPT
decrease
-71%
AI Overviews
increase
+611%
AI Mode
increase
+42%
Gemini
decrease
-35%
Perplexity
decrease
0%
Copilot
decrease
-64%
Decrease: ChatGPT, Gemini, Perplexity & Copilot
ChatGPT showed a large increase in AI citations for December and January, however this sudden increase corrected itself in February 2026 and in March saw a decrease. There is no isolated connection that Schema Markup directly caused an increase in AI visibility on ChatGPT.
Despite Gemini being the only platform that could fetch the Schema Markup when directly asked (see test 1). No significant increase in AI visibility was spotted after implementing structured data to the website.
Increase: Google AI Overviews & AI Mode
Google AI Mode, AI Overviews were the only platforms where we saw a gradual increase of AI citations during the time-period of the Schema GEO experiment. However, this the total AI mentions were still fewer citations than in August 2025.
We did manage to isolate and identify an uptick in both AI Overviews and classic SERP features for the entire website. However this increase was minimal at first.
Late February we saw an uptick on OtterlyAI but also on competitor brands in the same landscape, which indicates that this increase is likely due to algorithm changes inside Google.
So where does Google AI Overviews fit in (SEO, AEO, or GEO?)
Schema markup’s role here aligns with that: it contributes to SERP feature eligibility and indexing signals that feed into whether AI Overviews are triggered at all, which is why the +1,500% uplift reflects a traditional SEO mechanism, not a GEO one.
Once inside an AI Overview, however, the selection of which sources get cited behaves more like GEO. Content depth, entity clarity, and topical relevance determine inclusion, not ranking position alone. And the format of the answer itself, a direct, synthesized response to a specific query, is pure AEO. Google AI Overviews is therefore the one surface where all three disciplines converge, and the data from this experiment suggests the most actionable entry point remains the oldest one: solid technical SEO, with schema as part of that foundation.
What the data shows of the entire website:
OtterlyAI’s brand coverage increased approximately +6% in the weeks following December 7, moving from a November baseline of roughly 27% toward the low 30s. Viewed in isolation, this might appear to validate schema’s impact.
None of the competitors in the same space implemented schema changes in December. All of them saw equivalent or larger brand coverage movements during the same period.
The conclusion is direct: the December lift was algorithmic, not schema-driven. A platform-level change in how AI Search engines were weighting or retrieving sources in this category caused coverage to rise simultaneously across multiple brands. OtterlyAI’s schema implementation happened to coincide with that shift.
OtterlyAI’s brand coverage continued to grow through January and into February 2026, reaching approximately 49% by mid-March ,a trajectory consistent with ongoing content, citation building, and entity-building efforts across a broader GEO strategy, not attributable to any single technical implementation.
Test 3: Add Information That Can Only Found in Schema Markup
For the final test we added information that can only be found inside the schema and tested if the AI Search Platforms could answer these specific questions correctly.
Inside FAQ Schemas a question and answer was added that wasn’t listed anywhere else on the OtterlyAI site or external websites. The question was “How many OtterlyAI employees live in Portugal?”
Results
After several days, the results were already clear. None of the AI Search platforms used Schema Markup in an attempt to answer this question. Even when one of the prompt instructed to look at a very specific page:
Here’s an overview of how the AI Search Platforms tried to answer the question:
This third test is the final nail in the coffin of the theory that AI Search Platforms don’t prioritize Schema Markup. Our closing thoughts is that Schema Markup is still important, but should technically be seen as an SEO lever rather than a GEO lever.
What This Means for How You Approach Schema in GEO
Schema Is Not a Direct AI Citation Signal
Based on our testing on the OtterlyAI website, schema markup did not appear to directly influence AI Search citation behavior in a measurable way for most AI Search platforms. The direct fetch test showed that 6 out of 7 platforms could not access raw schema data at all, and the brand coverage lift we saw in December aligned closely with competitors who had made no schema changes, pointing to an algorithmic shift rather than a schema-driven effect.
This is not a reason to remove schema. It is a reason to stop treating it as a GEO growth lever and be honest about what it actually does.
The Indirect Path Still Exists: Schema → SEO → AI
Several AI Search platforms ,most notably Google AI Overviews ,use traditional organic search ranking signals as part of their source retrieval logic. Pages that rank well in organic search are more likely to appear in AI-generated answers for the same queries.
This creates an indirect pathway: schema markup → better rich results → stronger organic rankings → increased likelihood of AI citation.
The chain is real, but the mechanism is indirect. Schema contributes to AI visibility only insofar as it contributes to the traditional search performance that some AI platforms treat as a proxy for source quality. That’s very different from schema having a direct effect on how AI Search Platforms retrieve and cite content.
Platform Behavior Is Fragmented ,Including Schema Access
The fetch test results illustrate something that extends beyond schema: AI platforms process web content in fundamentally different ways. Gemini can read raw HTML structure. Most platforms convert pages to Markdown and lose structured data entirely. Google AI Mode generates plausible-sounding output regardless of what’s actually on the page.
This fragmentation affects every aspect of GEO strategy, not just schema. Any technical implementation that assumes a consistent retrieval layer across platforms is building on a false foundation. What is visible to Gemini may be invisible to ChatGPT or Perplexity.
The Schema Playbook for AI Search (The Honest Version)
1. Schema markup is an SEO lever, not a GEO growth lever. Schema remains a best practice for technical SEO and rich result eligibility. Maintain it. Audit it for accuracy. But don’t expect it to move your AI brand mention count or brand coverage score. The direct mechanism simply doesn’t exist for most AI platforms today.
2. Use schema to protect organic rankings ,that’s the indirect AI path. If your FAQ Schema earns a featured snippet, or your Article Schema contributes to better organic visibility, those rankings may eventually translate into AI citation inclusion ,because some AI platforms weight organic authority as a retrieval signal. Optimize schema for rich result eligibility, not for AI consumption directly.
3. Always control for competitors before attributing coverage changes. Any time you implement a technical change and see brand coverage move, check whether your tracked competitors moved in parallel. If they did, the cause is almost certainly a platform-level algorithmic shift. Treat isolated, single-brand lifts as the threshold for crediting a specific intervention.
4. Invest in what AI platforms can actually read: clear, structured prose. AI Search Platforms retrieve and process natural language. Strong entity clarity, explicit brand mentions in external citations, consistent terminology across the sources AI platforms already trust, and content written to answer the specific form of questions AI systems decompose ,these are far higher-leverage investments than JSON-LD annotations that most platforms strip before reading.
5. Gemini is the exception, but a limited one. Gemini is currently the only major AI Search platform that demonstrated an ability to read raw schema markup correctly. If Gemini represents a priority platform for your specific query category, schema may have marginal relevance there. But Gemini currently holds a very small share of AI Search citations in most tracked verticals. Optimizing primarily for Gemini’s schema-reading behavior is a low-ROI bet compared to building the external citation presence that drives visibility across all platforms.
6. For product discovery in AI Search, skip schema & apply for AI Merchant Programs. If product visibility inside AI Search is the goal, there is a more direct path than any technical markup: native merchant programs built directly into AI platforms.
ChatGPT launched OpenAI’s Merchant Program, allowing merchants to integrate live product feeds into ChatGPT Search results and enable Instant Checkout via the Agentic Commerce Protocol (ACP). Shopify and Etsy sellers are already eligible with no integration required. The program is live and accepts applications on a rolling basis.
Perplexity launched a similar Buy with Pro feature, enabling Pro subscribers to purchase products directly inside Perplexity responses via shoppable product cards ,this is one of the first AI Search platforms to move from citation to conversion natively.
Both programs bypass traditional SEO and GEO mechanics entirely. For product brands, applying is a higher-leverage move than any schema audit.
The bigger picture: AI agents will soon buy on your behalf. The ACP is the infrastructure layer for a fundamental shift in commerce. A user taps “Buy,” confirms payment ,and from that point, the AI handles everything: fulfillment options, sales tax, payment token, order confirmation.
AI won’t just recommend products, it will complete the purchase. Being discoverable inside AI Search is no longer just a traffic question, it is a transaction question.
Pro Tip: Discover our GEO experimentation sheet to track live updates on the latest AI Search studies
Closing Thoughts
Schema markup has earned its place in the technical SEO toolkit for a reason: after implementing it across the OtterlyAI website, we saw a +377% increase in SERP features and a +1,500% increase in AI Overviews appearances over three months. Those are meaningful results, and they make a strong case for keeping schema as a consistent part of your SEO practice.
What our experiment suggests is more specific: schema’s value appears to flow primarily through traditional SEO performance, not through direct consumption by most AI Search platforms. For the majority of platforms we tested, the structured annotation layer that schema provides was either inaccessible in the fetch process or didn’t appear to influence citation selection. This is based on our own website and our own data, your results may vary, and we’d encourage you to test this on your own site before drawing firm conclusions.
The bottom line: keep schema on your technical checklist, optimize it for SERP feature eligibility and rich result performance, and let those traditional SEO gains work as the indirect pathway into AI visibility. But if you’re prioritizing schema specifically as a GEO growth tactic, our experiment suggests that investment is better placed elsewhere.
For practitioners building GEO strategies: schema audits should stay on your technical checklist. They should come off your list of GEO growth priorities. The effort that goes into schema optimization delivers more AI visibility impact when redirected toward content clarity, external citation building, and the entity consistency that AI Search platforms actually use to decide which sources to trust.
If you want to track your own AI brand coverage and run controlled experiments like this, explore OtterlyAI’s AI Search Monitoring
Have You Run a Schema Experiment on Your Own Site?
The findings above are based entirely on OtterlyAI’s own website and data. Schema behavior may look different in ecommerce contexts, high-authority domains, or sites operating in different verticals. We’d genuinely love to hear what you’ve found.
Have you implemented schema markup and tracked the impact, either on SEO performance, AI Search visibility, or both? Did your results align with ours, or did you see something different? Share your findings with us. Real-world data from across different site types is how this field moves forward, and every experiment adds to the picture.
