That one sentence captures the pivot from classic Search Engine Optimisation (SEO) to Generative Experience Optimisation (GEO), the discipline of earning inclusion inside AI-generated answers. If your brand isn’t referenced when a buyer asks an assistant for shortlists, comparisons or recommendations, you’re simply not in the room.
SEO won the blue link. GEO wins the answer. Large language model (LLM) assistants and AI search layers assemble responses from multiple sources, official docs, trusted media, community threads and structured product data. The prize is no longer position 1. It’s to be cited as a reliable fact source or mentioned as a credible option in the response itself.
We’ve already seen the gatekeeper effect. A prospect asks an AI for a shortlist, and the assistants serve up brands they trust. As Tony Garner, Founder at Viva PR, put it, the game has changed because the machine now decides who gets invited.
“A prospect told us they’d put their brief into ChatGPT and asked: ‘Give me five agencies who could deliver this.’ And we were on the list. That was a ground shift. AIs are already acting as gatekeepers to business opportunities. The question is how do you unlock them? The answer is surprisingly familiar: it comes down to reputation and storytelling. Clear, well-structured owned content plus consistent earned mentions effectively ‘train’ the machine to understand who you are.”
That’s the strategic punchline. In GEO, your authority is assembled from what you say about yourself, and what the world says about you, with growing weight on the latter.
Classic SEO rewarded precise keyword targeting and on-page optimisation. GEO rewards coverage of user intent, often multi-step tasks framed as natural language questions. People don’t ask “best CRM pricing tiers” anymore; they ask, “I run a 15-person sales team — shortlist CRMs under £50 per seat with native WhatsApp support.” Assistants decompose that intent, then synthesise an answer and cite sources.
So the content brief evolves. You still need pages that rank, but you also need content that LLMs can easily parse, summarise and reuse. Think structured guides, comparison tables, FAQs, and product specs with crisp attributes, not prose-only marketing pages. As Sean Dwyer, Managing Director at Door4, notes, structure is now strategy.
“We’ve always said ‘write for humans’. That remains true. But a parable of prose won’t help an LLM extract your key facts. In GEO, structure and surfacing are the tough nut: schema, clear hierarchies, and machine-digestible product information so assistants can lift and cite with confidence.”
Backlinks won’t vanish, but their monopoly on authority has ended. Unlinked brand mentions—in reputable media, expert forums, and official documents—increasingly influence how AIs describe and include your brand. That’s uncomfortable for teams used to precise, measurable link-building, but it’s closer to how humans form trust.
“Traditional search engines used to mainly reward what you said about yourself. AI search now rewards what others say about you. I’ve heard that split quoted as about 20/80. That pivot turns PR from a predominantly Brand Marketing channel into a Performance Marketing channel. Earned media isn’t just influential, it’s algorithmically essential.”
Leon Calverley, Founder at Door4, sees a rhyme with early PageRank, but with a modern twist. “Citations matter,” Leon argues. “LLMs place value on real expertise demonstrated by brands and retailers. Links are likely less significant right now – but let’s be honest, nobody can prove the exact weighting.” That humility matters. We can observe patterns and outcomes, but signal weights are fluid and model-specific.
LLMs gravitate to authority. High-signal domains: established news outlets, Wikipedia, standards bodies, academic repositories, and your official product documentation are frequent citation targets. Assistants also incorporate licensed datasets, and contracts with platforms like Reddit have expanded access to community content. The implication is simple but demanding: become a source of record in your topic. That means:
As Tony frames it, PR storytelling now doubles as machine teaching. You’re not only persuading people; you’re shaping the training material that assistants will draw from.
Forums and social platforms increasingly appear in AI answers. Reddit threads, Stack Exchange, Quora discussions and LinkedIn posts offer grounded context, lived experience, and product comparisons that LLMs love to summarise. That raises the stakes for community visibility and reputation management.
Tom Morton, Head of Activation at Door4, is pragmatic: “These spaces will matter more because they already feed into training data. If people are talking about you positively in Reddit, Quora or LinkedIn, it strengthens trust signals. The label has changed, but the fundamentals haven’t: answer the user’s need.”
Be careful here. You can’t astroturf credibility. Community strategies that work in GEO are the same ones that work with actual humans: help first, sell later; bring evidence; show up consistently.
AI layers prefer current, well-maintained sources when the topic is time-sensitive; pricing, availability, compliance, or anything with a short half-life. If your site shows outdated specs, or your press page hasn’t been updated in 18 months, you reduce your chance of being cited. Build an editorial operating cadence: smaller, frequent updates beat sporadic big bangs.
Some publishers have reported steep declines in click-through from AI-augmented search surfaces. Whether those drops settle or rebound, the strategic response is the same: earn inclusion, not just clicks. The traffic model is evolving; your authority model should be too.
None of this replaces technical hygiene. Crawlability, mobile-first experiences, security and latency remain table stakes. If an assistant can’t reliably fetch and parse your content, it can’t cite you. Prioritise:
Sean’s question is a useful prompt for teams planning site changes: “Is this the real priority and hierarchy we wish to convey to bots?” If not, rewrite the hierarchy.
GEO demands new KPIs. Traditional ranking reports and organic sessions don’t capture whether assistants are naming or citing you. Consider a measurement stack that blends qualitative and quantitative signals:
How do you gather this? Start scrappily. Maintain a panel of priority tasks and prompts. Run them through leading assistants, record screenshots, and code the outcomes. Track branded and unbranded mentions across media and community. Build a simple scorecard that highlights gaps: missing price info, weak comparisons, no third-party validation, and outdated specs. It’s less elegant than a rank tracker, but closer to reality.
The PR–digital handshake is now non-negotiable. Assistants reward credible third-party corroboration, which is PR’s wheelhouse. But it’s more than placements; it’s narrative control and signal consistency.
“Put yourself in the position of an AI search bot, considering putting its neck on the line by recommending your business. Is it confident because your site has all the right keywords — or because it sees those claims echoed in trusted external media? In GEO, the link-up between digital teams and PR isn’t advisable; it’s a must.”
Tony agrees: traditional media will matter more in GEO because those outlets carry weight with both people and machines. That doesn’t diminish owned channels – it simply raises the bar. Your claims need receipts, preferably hosted somewhere other than your homepage.
If you’re short on time, here’s a focused plan to pivot from SEO to GEO without throwing out what already works:
Let’s keep our feet on the ground. LLMs are probabilistic. They hallucinate. They change quickly. No one outside the vendors can tell you the exact authority weights of links vs mentions vs recency – and those weights differ by model and by query. Leon’s caution stands: links may be less significant in practice today, but hard proof is elusive.
There are commercial trade-offs, too. AI answers can compress traffic to publishers and brand sites. You might get named but not clicked. That makes measurement messy and ROI attribution harder. But it doesn’t change the strategic necessity. If the answer is the new homepage, you need to appear in it, with language and claims you can live with.
Finally, there’s a governance question. If assistants are paraphrasing your claims, you want your source material to be consistent, attributable and safe. That means tighter internal standards on product facts, performance claims and comparative statements. PR, legal and product marketing need to sit closer together.
GEO doesn’t replace SEO; it reframes it. The mechanics change from keyword matches to intent coverage, from links to mentions, from clicks to citations, but the principle is familiar: become the best, most trusted answer to a specific need. Do that in public, in structured ways, and assistants will learn to recommend you.
One reflection to leave with: if your brand disappeared from AI answers tomorrow, which signals would be missing? The link, the mention, the proof, or the structure? Your next quarter’s plan sits somewhere in that list.
With thanks to our external contributors, Tony Garner and Greg Wilson, for their practical insights on how PR and earned authority now drive performance in the AI era.
Our latest articles, features and ramblings.
We explore performance marketing, AI, communications and optimisation.
