Skip to content
Menu
When digital PR first emerged around 2010, it meant social. It was a truly exciting time. I ran digital listening every day for the BBC. I helped introduce the first event Twitter customer assistants for Debenhams. And my live tweeting for Microsoft at industry events saw its ROI increase tenfold.
But as the decade progressed, digital PR teams turned their attention to SEO. It was admittedly less thrilling than the social innovations, but showing your client how much of page one of Google they were dominating because of your hard work was, and still is, extremely rewarding.
Now, with large language models (LLMs) shaping how millions of people discover brands, products, and reputations, digital PR is ready for another shift. If it were to rebrand its meaning once more, it would be all about Generative Engine Optimization (GEO).
At the recent PRmoment Masterclass on the intersection of PR and GEO, ten sessions explored what this shift means in practice, featuring speakers from all corners of the industry, including yours truly. The event made one thing clear: PR professionals who ignore GEO do so at their own risk, and, more importantly, the risk of their clients. Let’s dive into five key themes that stood out.
1. The audience has changed
PR has always been about reaching the right people with the right story. But content now needs to satisfy three distinct audiences simultaneously: journalists, stakeholders (customers, partners, shareholders etc.), and LLM algorithms. Each has different needs. Journalists want news value and credibility. Stakeholders want clarity and relevance. And AI models want structured, well-cited, authoritative information they can reference in their outputs.
This isn’t a small tweak to how campaigns are planned. It’s a fundamental shift in how content strategy needs to work. Brands that optimize for only one of these audiences will find themselves invisible to the others.
2. Visibility is an ecosystem, not a single channel
One of the strongest themes of the day was that LLM visibility isn’t built on any one type of content. It’s an ecosystem. At Fusion, our digital practice here at Fire on the Hill, our analysis of over two million AI citations across major LLMs, including Google AI Overview, ChatGPT, Perplexity, Claude, Gemini, and most recently CoPilot, has shown that the brands performing best are the ones with a fully rounded presence across five citation categories: owned corporate content, earned editorial, reference sites, user-generated content, and institutional sources.
Owned content and editorial are the foundations. Corporate pages tend to dominate branded queries, while editorial coverage leads for non-branded, category-level prompts. But brands that rely solely on those two, without investing in reference sites like Wikipedia and industry databases, UGC platforms like LinkedIn, Reddit, and Substack, or institutional sources like trade bodies and chartered organizations, consistently underperform compared to competitors who cover the full mix.
For B2B brands in particular, trade media over-indexes significantly compared to mainstream outlets. And within social, it’s LinkedIn articles specifically, not posts, that carry weight with LLMs. Reddit and Quora matter too, largely because the models appear to trust the community voting mechanisms. Meanwhile, Substack and Medium have become increasingly important as laid-off journalists from traditional outlets build their own subscriber bases, bringing their authority with them.
3. The story you place is not the story people receive
One of the most sobering themes of the day was the sheer breadth and depth of what AI models now draw on when they build a picture of your brand. The gap between the story you place and the story people actually receive has widened dramatically. In 2010, you placed a story and your audience read it. By 2020, algorithms decided who saw it. In 2026, your audience queries an AI model, which syntheizes a response from 50 or more sources, of which your carefully placed article is just one.
That can seem overwhelming. But consider what happens when someone prompts an AI model with a question about a brand’s reputation. The response does not come from a single article or press release. It pulls together financial data, operational changes, employee reviews, leadership moves, legacy controversies, and brand perception, citing dozens of sources from trade press to Glassdoor to social commentary. No single piece of coverage drives the narrative. The model builds its own. It draws on what might be called “structural signals”: the observable things an organization actually does. Hiring patterns, pricing changes, product decisions, supplier choices. These are signals that LLMs are getting very good at reading, and they matter far more than any press release.
These signals fall into three broad types: structural (what you do), consistency (whether your pricing matches your positioning, or your employer brand matches your Glassdoor), and authority (whether the people in your organization are making real, specific claims in their domain). If you are not proactively placing content in the sources that make up the ecosystem of a response, you lack control of what your story says. GEO is not optional.
4. Misinformation is free. Trust is expensive.
A recurring theme across the day was the role of misinformation in GEO. False or misleading information can spread rapidly through AI outputs, especially when models cite low-quality sources or aggregate unreliable data. The cost of correcting that misinformation, rebuilding trust, and ensuring accurate information dominates AI-generated answers, is significant.
This is where GEO becomes essential for brand protection. It is not enough to produce good content and hope for the best. Brands need to understand what AI models are saying about them right now, where those outputs are drawing their information from, and what needs to change to ensure accuracy.
At Fusion, our GEO service includes full auditing of AI chat responses to identify misinformation and put strategies in place to correct it. The speakers who addressed this topic were united on one point: the brands that invest in monitoring and managing their AI visibility today will be the ones that maintain trust with their audiences tomorrow.
5. GEO is already driving measurable results
The theory is compelling, but the numbers are already moving. Across the brands we work with at Fusion, we are seeing higher traffic from humans arriving via LLM referrals. We are seeing our clients’ coverage appear in LLM-cited results, increasing their share of voice against competitors. We are seeing overall visibility and position improve, while reducing the level of misinformation. The opportunity is growing fast, and the brands that invest in GEO now will capture a disproportionate share of it. Those that do not will find themselves increasingly invisible in the spaces where their stakeholders are looking for answers.
Of course, traditional search still dwarfs LLM traffic for most brands. SEO remains critical, and the good news is that strong SEO work tends to benefit GEO as well. They’re complementary. But the trajectory is clear, and GEO gives PR teams something they’ve always struggled to provide: direct, measurable proof that earned media activity drives traffic, leads, and, ultimately, revenue.
What comes next…
GEO is not a passing trend. It’s a structural change in how information is discovered and consumed, and it’s accelerating. The PRmoment Masterclass brought together practitioners who are already working at this intersection, and the consensus was clear: PR teams that integrate GEO into their thinking now will be the ones leading the conversation in twelve months’ time.
For brands, the message is equally straightforward. Your reputation in the AI era isn’t just shaped by what you say. It’s shaped by what AI says about you.
If you want to understand what AI is saying about your brand, and what to do about it, get in touch with our Fusion team.
