From SEO to GEO: How agencies are navigating LLM-driven search

The bread and butter of Google’s rankings have always been website links. Over time, it expanded results to include links to video, images, the latest news, forum posts, shopping sites and more for a comprehensive view of a search query by category.

Then came “knowledge panels,” pop-up boxes offering curated information at a glance.

Now, unless you’ve been on a year-long digital detox, you've noticed something bigger: Google is leading many search results with AI-generated summaries, powered by its large language model, Gemini. 

The summaries don’t just provide quick answers; they add context in conversational language, compiling information in a way that makes clicking on links feel increasingly optional.

Users don’t even need to go to a traditional search engine. In October 2024, OpenAI’s ChatGTP introduced a search function.

But whether ChatGTP or Google, AI search is quickly gaining favor. On Wednesday, Joanna Stern, The Wall Street Journal’s senior personal technology columnist, wrote, “ads and search-optimized junk made a mess of the go-to engine. Now ChatGPT, Perplexity, Claude — and even Google’s own AI — do it better.” 

The headline on her column reads, “I quit Google Search for AI — and I’m not going back.”

She’s not the only one. A Vox Media study in December found 61% of Gen Z and 53% of millennials turn to AI tools instead of Google or other traditional search engines to find information. This trend is also reflected in market movement. Google’s global search market share dropped below 90% for the first time since 2015 in the last three months of 2024, according to StatCounter. 

Where does the integration of AI into search leave earned media? Will traffic to media sites free-fall? What about earned media in the role of SEO strategies? 

Tejas Totade, chief technology officer at Ruder Finn, says search is at a “seminal moment.” 

“Search used to be considered a gateway to content. But it has become the end destination, which is the biggest shift in how we’re thinking about search at Ruder Finn,” he says. “As communicators and marketers, our focus had always been, ‘How do we show up positively on the first page of Google?’ Now it’s not just about showing up; it’s also about understanding what the LLM is going to say about you, because of how prescriptive it is.”

Add to that the conversational delivery, and it’s easy to see why someone might take whatever the AI generates at face value. This has massive implications for the proliferation of misinformation or disinformation. 

LLM search could also diminish the role of the byline and even the media brand as a marker of trustworthiness. 

“In the age of LLM-driven results, if the byline doesn't appear, does a journalist's credibility still matter?” Totade asks rhetorically. “Of course it does in the grand scheme, but from a user’s perspective, everything starts to feel like just an API — one massive RSS feed pulled into a tiny window.”

Whether it poses an existential threat or not to the media, Totade says that what’s clear is “content is becoming even more commoditized. It’s still king, but perceptions are being shaped by what appears in AI results.”  

In November, Ruder Finn launched rf.aio, an LLM-optimization service, which includes monitoring of ChatGTP, Anthropic’s Claude, Google’s Gemini and Meta’s Llama to evaluate how they respond to brand-related queries. 

Learning from monitoring will have a major impact on brands’ earned media strategy. 

“We’re already finding unexpected differences in citations from AI,” he says. “You might think, for example, a story on how to maximize credit-card points to book your next Bali vacation might cite, say, the Financial Times or Wall Street Journal. But guess what? It’s probably going to come from NerdWallet.” 

Why? That’s because AI systems tend to favor content that resonates with certain audiences, like millennials, who love practical advice, and that’s the type of content NerdWallet specializes in. 

AI platforms have also been signing licensing deals that will influence summaries and citations. ChatGTP, for instance, has partnerships with the likes of Conde Nast, the Associated Press and the Financial Times, the latter which led to concerns from PR pros in the U.K. 

“What does this mean for newspaper paywalls? Why would you continue to pay for a subscription for the newspaper if you can just get the info from AI?” posted Flo Powell, joint MD at Midnight Communications, a B2B PR agency. 

Todd Ringler, head of U.S. media at Edelman, says, “so-called generative engine optimization is going to be front-and-center in any successful brand or reputation campaign.”

“Unlike SEO,” he says, “GEO focuses on authoritative content to give it a leg up on discoverability within AI platforms. Among other changes, multimedia integration and earned media stories, including those behind the paywall, take on new currency in this changing value equation. Earned media and content strategies need to be savvy to where and how AI search is finding and structuring its answers and actively work to shape the information.”

In May, Edelman is “launching a new offering that analyses these dynamics” aimed at helping companies define critical swim lanes for their brands. “It’s an exciting evolution and layering of our current media strategies,” adds Ringler.

LLM search could also increase the prevalence of disinformation and misinformation, if fewer people dig deeper than AI summaries.

“This is something clients will need to closely monitor and act fast on, before misinformation and disinformation gets picked up by AI search,” says Christopher Hippolyte, deputy head of the corporate affairs practice at Syneos Health. 

Since LLMs appear to be prioritizing content from authoritative voices, “a part of your earned media strategy in healthcare needs to be engaging multiple stakeholders, [such as] healthcare advocates and even scientific organizations and nonprofits, such as the American Medical Association or American Association for Cancer Research.”  

"While still in its early stages, GEO is evolving quickly as a discipline and making waves,” says Charlie Baldwin, SVP of insights and analytics at We. Communications, noting “organizations worldwide are busy developing theories as they work to reverse-engineer LLMs.” 

He says it’s “all in an effort to figure out how to get their content noticed at the top.”

We. is conducting its own analysis to deepen clients’ understanding. “What is abundantly clear to us already is that telling a strong and clearly worded story is more important than ever,” Baldwin says. “Appearing in credible, authoritative publications and focusing on content that offers authentic, ownable language is essential.” 

Distinct language that stands out could become a competitive advantage for smaller brands. 

“While this approach is critically important for market leaders, it's also a fantastic opportunity for challenger brands,” he says. “A distinctive, niche positioning hasn't always been an advantage in search, but as people increasingly turn to AI for specific recommendations, brands that used to be buried in traditional search results could be surfaced more prominently.”

Despite the emphasis on GEO, brands still need to get their messaging across in longer storytelling.  

“Optimizing for GEO helps, but we still need to make content worth reading in full so that the message can be delivered in the full brand context, be it a story or information page,” says Ephraim Cohen, global head of data and digital at FleishmanHillard. “I refer to it as the continually evolving CliffsNotes challenge: Is it enough to read a summary of Frankenstein? Or is it worth reading the full book? Content needs to be worthwhile for the right audience.”

“Like all cycles, the challenge is figuring out what type of content is valuable enough for people to click through to and read — and, in paywall cases, pay for — versus reading what is essentially a summary or a custom article in the form of a generative AI answer,” Cohen emphasizes. 

As for where search is headed, he sees a couple scenarios. 

On one end of the spectrum, AI search could actually boost some earned media traffic with smarter summaries that drive curiosity.

“People will want to click through for the full context or deeper story,” he says. That’s what we see today in the volume of GenAI traffic versus searches that are clicked through. In that scenario, earned media can thrive, especially if outlets make sure their stories are clearly structured and easy to pull insights from — similar to SEO best practices.”

But there is also another scenario.  

“If AI responses become so good that people rarely click through, content creators will shift more toward being information providers to the machines. The challenge becomes, how do you ensure your voice, message and facts are what’s getting pulled and presented? That’s a new layer of strategy, less about optimizing for search engines, more about optimizing for AI training and outputs,” says Cohen. “There are similarities to social media, where people were looking at social content more for news versus reading original news sources. The result of that was we started treating the right content creators as news sources.”

“This is an area we are exploring with intention now: forming a capability around optimizing for AI training and outputs,” he says. 

He says new solutions could emerge, as well. LLMs, for instance, could end up working with agencies. 

“We can then aggregate our brands’ trusted content to feed directly for training models similar to how they are cutting deals with other content providers such as news organizations,” says Cohen.