GEO, SEO, and the Messy Reality of AI Visibility 

By Ben Dillon, CEO

The work to influence those answers goes by a rotating cast of acronyms: LLMO, AEO, GEO. At Geonetric, we’ve landed on GEO (Generative Engine Optimization), and I’ll use that here. 

But after a few days of panels and hallway conversations, I’m convinced of something uncomfortable: while everyone is talking about GEO, a lot of what’s being recommended right now is either incomplete, overconfident, or aimed at the wrong problems. 

Here’s what’s actually worth paying attention to. 

Technical SEO is back, whether you like it or not 

For the last several years, SEO strategy has mostly meant content strategy. What themes are we covering? What questions are we answering? How much can we publish? 

That still matters. But GEO is dragging something out of the neglected corners of healthcare marketing programs: technical SEO. 

Schema markup that is systematic, not one-off. Structured data that builds an actual knowledge graph. Content organized explicitly for machine interpretation, not just human consumption. These are the things most teams have acknowledged for years and quietly deprioritized. 

GEO for healthcare websites changes that calculus. 

If an AI cannot reliably understand who you are, what you do, which physicians work for you, what conditions you treat, and where you operate, you do not get weighed. You simply disappear. 

This is not about ranking better. It is about being legible at all. 

The “credibility” framing is comforting and slightly misleading 

Kirsten Lecky from WG Content and Robbie Schneider from Franciscan Health presented research at Swaay.Health on AI visibility in healthcare, supported by WG Content’s broader testing across 15 health systems. 

The research is solid, and the recommendations are directionally right. Where I part ways slightly is with how much of this gets framed as a question of “credibility.” 

That word implies judgment. It implies a consistent preference order. I don’t think that’s what’s happening. 

What seems closer to reality is what Rand Fishkin has described: large language models are pattern-matching across an enormous and messy corpus. They are not scoring you against a fixed rubric. They are assembling a plausible answer from probabilistic associations. 

That distinction matters. 

Ask an AI who the best cardiologists in your city are ten times, and you will almost certainly get ten different lists. Different organizations. Different providers. Different ordering. That inconsistency is not a flaw. It’s the system working as designed.  It’s inherent in the way that LLMs work. 

If you expect GEO to behave like SEO with a new label, you will invest in the wrong things and misread the results. 

Your website is still the main character, just not the only character 

One of the strongest signals in the WG Content research, and one that many marketers are still underestimating, is how much the external web shapes AI responses. 

In traditional SEO, your website is the primary object of optimization. In GEO, it’s your brand. 

News coverage. Reddit threads. Glassdoor reviews. Patient ratings. Social posts. Industry commentary. The AI pulls from all of it, weighs it unevenly, and synthesizes an answer you do not control. 

That makes a lot of healthcare marketing leaders uncomfortable, especially those who have spent years carefully shaping on-site messaging. But there is no opting out. 

GEO turns reputation management, media relations, and online sentiment from “nice-to-have” PR functions into core marketing infrastructure. If you are still treating them as adjacent concerns, AI will make that gap visible very quickly. 

No, SEO is not dead 

There is a growing narrative that GEO makes SEO, or even your website, irrelevant. That story spreads well. It is also wrong. 

LLMs are increasingly doing live web searches, pulling current pages, and synthesizing what they find. If you are not discoverable through traditional search, you are often invisible to the AI as well. 

Falling out of SEO visibility shrinks the pool of evidence the AI can draw from. Shrink the pool enough, and you disappear from the answer entirely. 

GEO does not replace SEO. It depends on it. 

The ranking question is the trap 

Where skepticism is absolutely warranted is around measurement. 

Rand Fishkin and the SparkToro team ran the same prompts hundreds of times across ChatGPT, Claude, and Google AI. The outcome should make anyone selling AI rank tracking very nervous. 

In most cases, there was less than a one percent chance of getting the same list of recommended brands in two runs. The idea of a stable, reportable “AI ranking” is far less real than many dashboards imply. 

Everyone wants a simple answer to “how do we rank in AI.” Right now, the honest answer is: not in a way that maps cleanly to traditional metrics. Pretending otherwise creates a false sense of precision. 

The market may be overcorrecting 

Here’s a preview from Geonetric’s upcoming 2026 Healthcare Marketing Trends research. When we ask health system marketers to rate the importance of different channels, the perceived importance of SEO dropped sharply this year. 

I understand the instinct. GEO feels new and strategic. SEO feels old and operational. 

But that is an overcorrection. 

The organizations that will outperform are not choosing between SEO and GEO. They are maintaining strong search fundamentals, strengthening their technical foundation, and deliberately expanding their thinking to account for how AI actually builds answers. 

Treating this as an either-or decision is how teams create messes they spend the next five years untangling. 

If you want the full data when we publish the 2026 Healthcare Digital Marketing Survey, get on the waitlist.