In this video, Etna Interactive CEO Ryan Miller, discusses how to monitor your practice’s visibility in AI Search. As patients increasingly turn to tools like ChatGPT, Google’s AI Overviews, Claude, and Perplexity for provider recommendations, tracking “share of voice” has become just as important as monitoring traditional search rankings.
Follow along to learn the difference between organic search rankings and AI search monitoring, which four engines you should be sampling, how to build a prompt rubric anchored around your key procedures and topics, and how often you should be checking your results to get a true, representative picture of your practice’s AI search performance.
Video Transcription:
Ryan Miller: Hi again, Ryan Miller here with Etna Interactive, and we’re going to talk a little bit today about monitoring your positioning in AI Search.
Now, we’re recording in July of 2026, and you might be surprised to learn that AI systems like ChatGPT, Google’s AI Overview, Claude, Perplexity are now the third most popular online channel for discovering local businesses.
Put simply, patients are using AI Search for provider recommendations. And if all you have is anecdotal awareness of how often you’re being cited in AI search, you have a real, immediate, and existential challenge because you lack the insight you need to make critical business decisions about how you’re found now and going into the future.
Now, we’re going to solve that today, and I want to start by talking about the difference between monitoring traditional organic search and monitoring how you’re showing up in AI search systems.
Now, recognize that Google ranked individual pages where AI systems are synthesizing answers back to consumers. They’re not doing that for a small, fixed set of popular keywords. They’re doing it across a broad range of natural language prompts or queries.
Now, we have to shift our thinking from individual keyword rankings to a new metric called share of voice. Now, this has two different dimensions that are baked into it.
The first is how often you show up, how often you are mentioned or cited by the AI system, and how prominently you show up within those citations.
But that’s not it. There’s a new third dimension as well, where we have to think about sentiment, because AI systems are not just ranking sites, they’re characterizing practices.
And I’ll use this example.
Look in the lower left side of your screen. I want to circle a couple of things there for you. I asked for a facelift surgeon recommendation, and importantly, right away, ChatGPT came back and said, “Here are the most respected”—we would never see language like that on a traditional Google Organic Search: Surgeons who are known for very specific things and who have very specific, characterized reputations, right?
So, we have to recognize that we have to monitor both visibility and sentiment. So, let’s break that down.
Well, the first question I get asked is, “Where do I need to worry about monitoring my visibility?” And we recommend that you sample at least 4 engines.
Google—with their AI overviews and their Gemini platform—they have the broadest reach today, because consumers are still often starting their search journey on Google. But the most popular, independent, AI-only system that people are searching is going to be ChatGPT. So, those are our first 2.
We recommend keeping Copilot in the mix because people who are Microsoft loyalists are going to naturally be exposed to the product. And importantly, there’s a few smaller, AI systems like Claude and Perplexity that can round out and give you a representative sampling.
It’s not just about sampling multiple engines, though, because there can be a high degree of variability across the different AI systems. It’s also sampling using multiple prompts.
One query is going to show you one possible answer, and that’s not sufficient because patients are phrasing the way that they search very differently every single time. We want broad samples that give us patterns, that deep understanding of share of voice, not an anecdotal point sample, right? Not an expression of whether you beat the luck of the draw on that moment in time.
So, how do we arrive at the sampling? Now, we’ve carefully tested and developed a rubric that guides the prompts that we use for our AI search monitoring.
But for everyone who’s not an Etna client, here’s a little background to help you out. There is no public database like there is from Google today that shows which queries are most popular.
And, realistically, we don’t expect that to come any time soon. Google started sharing that data historically only after their ads platform became popular, and they realized sharing keyword search volumes would help people spend more money in their advertising platform.
We recommend, instead, that you anchor your queries around your top topics. These will include your category, like plastic surgery or medical spa, and the procedures for which you most wish to be known.
Now, importantly, there’s a lot of types of queries that you can, prompts that you could develop. We recommend prompts that return brand citations. And so, this would be prioritizing something like, “Who are the leading facelift surgeons in Tucson?”
And a little less emphasis today, unless strategically, this is important to you, on being mentioned in “How long is facelift recovery?”
The former is going to return a specific list of recommended practices, a priority we would make the case for practices today. The latter is going to return a synthesized answer with some abstract citations about which websites influence the answer that are far less likely to get clicked and result in immediate new business opportunities.
Now, importantly, once that’s done, you need phrase variants in all of your prompts to reflect the natural conversations that people are going to have with AI systems. We recommend a minimum of 5 for every topic.
The next question you might ask is, well, “When do I need to do this sampling?” And our recommendation specifically is a minimum of twice a month, and that you maintain that cadence over time because snapshots can be misleading.
AI systems are evolving each and every day. You ask the same query 2 times on the same day [on] the same AI search engines, and you can get very different results.
So, sampling is going to give you that representative understanding of how your share of voice is evolving, and consistency is going to allow you to be a good steward, to make smart moves in terms of influencing your performance across the range of queries, the range of search engines over the long haul. So, as a business leader, here’s what we want you to know.
It’s important that you draw recurring samples, multiple sample points every month, for a representative set of prompts. So, these are representative across the topics that you want to be known for and natural language that will return specifically branded citations or practice recommendations, and importantly, that you’re doing that across the 4 leading AI search engines.
As a clinic leader, this is going to be the insight you need to make smart moves on where and how you can grow your practice with AI Search.
Leave a Comment