Google now shows two different answer layers on the same search results page, and they're pulling from almost entirely different sources. A new benchmark study from B2B SaaS agency DerivateX analyzed over 1,200 AI Overview citations and found that only 35% of the sources Google's AI Overview cites also rank in Google's own top 10 organic search results for the same query. The remaining 65% appear exclusively in the AI Overview.

The gap widens dramatically when it comes to product recommendations, which is where most buyers actually make decisions. Only 28% of the products named in Google's AI Overview also appear in the traditional search results below it. In other words, 72% of the products the AI recommends never show up in the classic blue-link search results at all.

Why Are Google's Two Search Layers So Different?

The study, titled "Two Googles, One Query," examined 100 buyer-intent software queries across 20 product categories, capturing data in June and July 2026. Most people assume the AI Overview is simply a summary of the search results beneath it, but the data tells a different story. Google's AI systems appear to have their own content preferences that diverge significantly from traditional ranking signals.

One striking pattern emerged: YouTube videos appear in 51 of the 100 AI Overview responses, but Google's traditional search results rank YouTube in only 7 of those same queries. Reddit shows the opposite pattern, ranking in 82 of 100 search results but cited by the AI Overview in just 43. This suggests that generative AI systems and traditional search algorithms weight different types of content very differently.

According to Google's own guidance on optimizing for generative AI features, the company uses a process called retrieval-augmented generation (RAG), which retrieves relevant pages from its search index and then uses that information to create AI-powered answers. Google also employs something called query fan-out, meaning its systems may generate several related searches in the background to better understand complex queries. However, the DerivateX research suggests these processes are producing distinctly different results than traditional ranking.

What Does This Mean for Websites and Content Creators?

The implications are significant for anyone trying to get discovered online. Ranking well in traditional Google search is no longer sufficient to appear in AI Overviews.

"Ranking #1 on Google still matters. It just no longer matters for the same reason it did in 2023," stated Apoorv Sharma, co-founder of DerivateX. "The AI Overview is a separate discovery layer with its own content preferences. Treating it as a byproduct of good SEO is the fastest way to disappear from the surface most buyers now see first."

Apoorv Sharma, Co-founder at DerivateX

The research identified three key structural gaps between the two discovery surfaces. When the AI Overview and search results do agree on sources, they tend to agree at the top; 72% of shared sources rank in Google's top 5, with a median position of #4. However, ranking well is not sufficient to guarantee AI citation.

How to Optimize Content for AI Search Engines

• Create Non-Commodity Content: Google emphasizes that generative AI systems can already summarize basic information very well, so generic content that repeats what dozens of other websites have said offers little value. Instead, create original research, expert commentary, real case studies, product comparisons, and lessons learned from actual client work that readers cannot find elsewhere.
• Treat Video as a First-Class Surface: The study found that video is the single largest structural gap between AI Overviews and traditional search results. YouTube appears far more frequently in AI responses than in organic rankings, suggesting video content deserves dedicated optimization effort.
• Earn Placement in Third-Party Listicles: The AI Overview cites third-party listicles and roundup articles in 63.4% of its citations, making these sources disproportionately influential for AI discovery. Getting featured in industry roundups and comparison articles is now a critical visibility strategy.
• Organize Content for Clarity: Use clear headings, short paragraphs, descriptive subheadings, and logical sections. Google recommends writing for people first and making pages easy to follow, rather than creating special AI-friendly formats.
• Support Written Content with High-Quality Visuals: Include helpful images, screenshots, product demos, explainer graphics, charts, comparison tables, and short videos where they genuinely help readers understand something faster or better. Follow image and video SEO best practices like using descriptive file names and adding helpful alt text.

The broader message from Google's own guidance is that SEO remains foundational for AI search success. As Brendon Kraham, VP of Search and Commerce Global Ads Solutions at Google, noted: "Your existing investment in solid, foundational SEO is your launchpad for AI success. That's why good SEO is good GEO (or AEO, or AI SEO, or whatever)". GEO stands for generative engine optimization, and AEO stands for answer engine optimization, both emerging disciplines focused on visibility in AI-powered search experiences.

Brendon Kraham, VP of Search and Commerce Global Ads Solutions at Google

However, the DerivateX research suggests that simply doing traditional SEO better is no longer enough. Websites need to measure AI Overview citations and Google rankings as two separate metrics rather than assuming one predicts the other. The study found that of the 20 product categories tested, 19 diverged significantly between the two surfaces. Only QuickBooks hosting showed close alignment, with 62% overlap.

For B2B SaaS companies and content creators, the takeaway is clear: the search landscape has fundamentally shifted. There is no longer a single "Google" to optimize for. Instead, there are now multiple discovery layers, each with its own content preferences and citation patterns. Success requires understanding and optimizing for each surface separately, rather than assuming that traditional search optimization will automatically translate to AI visibility.