- You cannot manage what you cannot see: most practices have never checked what AI assistants say when a patient asks about their specialty.
- A GEO audit runs real patient questions across ChatGPT, Perplexity, Gemini, and Google AI Overviews and logs whether you are recommended, mentioned, absent, or misdescribed.
- The four failure modes — invisible, misattributed, outdated, and ceded-to-aggregators — each have a different fix.
- Treat the audit as a standing instrument, not a one-time check. AI answers move week to week; your visibility is a metric, not a milestone.
Ask ChatGPT to recommend a dermatologist in your city. Ask Perplexity whether your flagship procedure is safe. Ask Google the same question a nervous parent would type at midnight. Then read the answer the way your next patient will: as advice, not as a search result. For most healthcare brands, that is the first time anyone has looked — and the answer is rarely the one they would have written.
Why the query layer moved
For two decades, being found meant ranking on Google. The patient typed, scanned ten blue links, and chose. That funnel still exists, but a growing share of it now terminates one step earlier: the patient asks an assistant, reads a synthesized answer, and acts on it. The assistant has already done the choosing. Gartner projects that traditional search-engine volume will fall roughly 25% by 2026 as users shift to AI chatbots and virtual agents.
This changes the unit of visibility. You are no longer competing for a rank; you are competing to be the entity the model names when it answers a human question. If it does not name you, there is no second page to be on. You are simply absent from the conversation that decided where the patient went.
What a GEO audit actually is
A generative engine optimization audit is disciplined, not magical. You assemble the real questions patients ask — not keywords, but full natural-language prompts: “Who are the best-reviewed knee surgeons near me?”, “Is this clinic legitimate?”, “What does this treatment cost and is it covered?” You run each one across the assistants your patients actually use, and you log four things: whether your brand appears, in what role, how accurately it is described, and which sources the model cited to get there.
That last column is the one most teams skip and the one that matters most. The model did not invent its answer; it assembled it from sources it trusts. The audit's real output is not a score — it is the map of which third-party pages are speaking on your behalf, and what they are saying.
Reading the results: four failure modes
Invisible. The model names competitors and never mentions you. Usually a corroboration problem: not enough independent, structured sources affirm who you are and what you do.
Misattributed. You appear, but the model assigns you the wrong specialty, location, or affiliation. A data-consistency problem across your site, directories, and profiles.
Outdated. The model describes a version of you from two years ago — a closed location, a former name, a service you no longer offer. A freshness problem.
Ceded to aggregators. The model answers entirely through directories and review platforms, and you are a line item inside someone else's brand. A demand-ownership problem.
Fixing what the audit finds
Each failure mode has a different lever. Invisibility is fixed by giving the models more to corroborate: structured facts on your own pages, consistent presence across the sources they cite, and content that answers the question directly rather than performing for a keyword. Misattribution and staleness are fixed by treating your entity data — name, locations, specialties, providers — as a single source of truth and propagating it everywhere the models read.
None of this is a one-time project. The models retrain, re-rank their sources, and change how they synthesize. A page that earned you a recommendation in March can quietly drop out by June with no signal to you at all.
The 1ness Take
Most practices discover their GEO position by accident, usually when a patient repeats something an AI told them that was wrong. By then the answer has been shaping referrals for months. The fix is not a clever prompt or a one-off agency report; it is making the audit a standing instrument — the same real questions, the same engines, run on a cadence, with the results tracked like any other acquisition metric.
Visibility in AI answers is not a milestone you reach. It is a number that moves, and the brands that win are the ones watching it on purpose.
