Most teams think they have an AI visibility problem. In practice, what they’re facing is something more structural: fragmentation.
New data shows the same brand can see citation volumes differ by as much as 615x between AI platforms, while overlap between top-cited sources in Google AI Overviews sits at roughly 17%. That level of divergence signals a deeper issue. What looks like a single emerging channel is, in reality, a collection of fundamentally different systems operating in parallel.
Despite that, many organizations are still approaching AI search as if it were unified—something you can optimize for once and scale across environments. That assumption is increasingly difficult to sustain.
Shane H. Tepper, co-founder of Resonate Labs, a company that helps B2B businesses be found and cited in AI search models like ChatGPT, Perplexity and Gemini, has been working with teams trying to map how brands actually appear across these systems. What they’re finding is not a visibility gap in the traditional sense, but a lack of transferability. Performance in one platform does not reliably carry over to another.
For years, search operated on a relatively coherent logic. Rankings fluctuated, but the underlying system was shared. Strong performance in Google translated into broader discoverability. AI search does not follow that pattern. Platforms like ChatGPT, Perplexity, Gemini and Google AI Overviews are not simply different interfaces layered on top of the same index. They retrieve, interpret and assemble information using distinct mechanisms, each with its own biases and dependencies.
The consequence is a landscape where visibility becomes fragmented. A brand that appears consistently in one system can be entirely absent in another—not buried on page two, but missing from the answer altogether. This is where many current strategies begin to break down. Teams validate visibility in a single environment, often through prompt testing, and extrapolate that performance across the category. In reality, they are optimizing for a system that does not exist.
This fragmentation introduces a less obvious risk: false confidence. When a brand surfaces in AI-generated responses, it can create the impression of coverage. But that visibility is often localized. It reflects how one platform retrieves information, not how the broader ecosystem behaves. As a result, what appears to be strong performance may in fact be partial exposure.
Partial exposure does not compound in the same way. It does not reinforce familiarity across multiple touchpoints or consistently position a brand during early-stage research. Instead, it produces uneven demand capture, where some buyers encounter the brand at the right moment while others never see it at all. The impact is difficult to measure because it doesn’t show up as a clear decline. It shows up as missed consideration that never enters the funnel.
The reason this happens is tied to how each platform is constructed. Google’s AI Overviews still lean heavily on its search index, favoring content that performs well in traditional rankings and incorporating first-party properties such as YouTube. ChatGPT prioritizes structured, extractable information—content that can be clearly interpreted and recombined into responses. Perplexity tends to reward breadth and corroboration, synthesizing across multiple sources. Gemini, while conversational in interface, remains closely aligned with Google’s broader ecosystem.
These differences are not superficial. They determine what gets retrieved, what gets cited and ultimately what shapes the user’s perception. There is no single optimization layer that sits above them.
This shifts the strategic question. It is no longer enough to ask how a brand ranks or whether it appears in AI-generated answers in general. The more relevant question is where it appears—and where it does not. In a fragmented system, presence is distributed. A brand can be highly visible in one environment, competitive in another and effectively invisible in a third, all at the same time.
Treating AI search as a unified channel compresses that reality into a single metric, masking the gaps that matter most. It also obscures how buyers actually behave. Users are not confined to one platform. They move between systems, often within the same research journey, comparing outputs and forming impressions based on what each environment presents.
For teams trying to adapt, the implication is not that they need to optimize everywhere at once, but that they need to understand where they currently stand in each environment. Visibility has to be measured at the platform level, and performance needs to be interpreted in context rather than averaged into a single score.
The broader shift is not just technical. It is conceptual. AI search does not represent a new version of the same system. It represents a fragmented layer of discovery where influence is distributed across multiple, non-aligned environments.
In that context, the risk is not simply low visibility. It is the illusion of visibility. And for many brands, that illusion is already shaping decisions they never get the chance to see.