For most of this century, healthcare providers have built their digital marketing plans to focus on chasing Google’s algorithm to squeeze every ounce of SEO juice out of a page. But the game is changing.
AI-powered models like ChatGPT, Perplexity and Claude are now answering people’s questions directly … and people like it.
Over the last year or so, just about every company has seen its website traffic fall off a cliff. This is largely driven by the adoption of AI—people can get answers without having to visit your website, whether Google is serving up an AI answer or they’re using ChatGPT, Perplexity or another large language model (LLM). According to a recent survey, 39 percent of respondents said they trusted tools such as ChatGPT to navigate healthcare decisions. Simply put, the responses from LLMs may not be completely reliable, but busy would-be patients rationalize that it beats an hour and $100 at a doctor’s office.
Welcome to the future—where the way your brand is discovered is no longer about tweaking a title tag or stuffing in keywords. It’s about what you know and how clearly you can explain it to both humans and machines.
Plenty of SEO professionals are trying to convince folks (and maybe themselves) that this is just a slight change in direction. But it isn’t.
The dawning era of generative engine optimization (GEO) is not a small tweak to SEO; it’s a fundamentally different approach to having your brand discovered. There are two primary reasons for this.
1. LLMs are simply a better experience for the user
Human beings, including your patients, are moving to LLMs for answers to their questions because it is a far superior experience—you get an answer instead of some half-hearted suggestions from a search engine (eight of which it’s being paid to tell you about). This saves patients time and money.
2. LLMs search for meaning rather than counting words
Where search engines have historically been math-oriented, tallying up the number of keywords and backlinks, they are increasingly focused on understanding intent and semantic meaning. But LLMs are far more nuanced in this regard—they are language-oriented, seeking out the meaning of the content
Google created algorithms like PageRank to gauge authority. LLMs come at this from the complete opposite angle—rewarding content from what the AI considers credible sources. But the credibility is not assigned by some made-up metric; AI determines credibility in a far more human way; they “train” on data that provides clear signals of expertise, credibility and influence within a particular topic or domain, and this influences how they answer your prompt.
How does your company demonstrate expertise to LLMs?
Human beings—and LLMs—understand that true experts can talk about a given topic for hours. They can write about it for dozens and dozens of pages and understand the hierarchy of what is important, so they can provide context and understanding. This is what LLMs crave—clear and understandable explanations of complex concepts.
The LLMs will also reward you for being able to provide breadth and context around a topic. Context creates understanding, so demonstrating how tangential topics, concepts and entities fit together is incredibly valuable. Make sure that content is structured in a way that appeals to LLMs. LLMs do not consume entire web pages; rather, it’s best to think of them as consuming content in “chunks,” and therefore to make those chunks as complete and clearly written as possible.
When in doubt, focus on creating content that human beings value—smart, differentiated, complete and concise, especially in such a complex field as healthcare.
Stop clinging to the SEO mindset
There is some overlap in how search engines and LLMs work, but the differences far outweigh the similarities.
Some LLMs generate answers based on pre-existing knowledge that they’ve been trained on, not by querying live web pages. They are trained on diverse datasets from the internet and beyond, identifying patterns of information and credibility based on the data they’ve been exposed to during training. This training process gives more weight to information from established, credible sources—often including the media. When you get your thought leadership published by a trusted gatekeeper like an editor, you’re demonstrating expertise, and that credibility carries weight with AI models (and human beings!).
A lot of SEO professionals say that optimizing for large language models is just a small deviation from what they’ve always done, but the evidence suggests this just isn’t true.
LLMs find information in a fundamentally different way compared to search engines. This chart explains some of the differences.
It’s a monumental difference from traditional SEO, and the biggest reason that SEO professionals do not have the toolkit to bring you into this emerging era. “Content” should not be about words in a Google doc—it should be focused on lucidly explaining complex ideas, i.e. traditional thought leadership.
Relying on SEO alone in an AI-first world is like optimizing for yesterday’s sunset.
Welcome to a brand new day.
John Miller
John Miller is the Founder and President of Scribewise, and the author of "Playing it Safe Sucks; A Manifesto for Courageous Marketing." Prior to founding Scribewise in 2012, he spent over two decades in numerous newsrooms and public relations roles, ultimately becoming a partner at one of Philadelphia’s most successful PR agencies.
