Every time search changes, the same people appear.
You know them. They turn up on LinkedIn the moment Google does something interesting, their post already written, their take already sanded smooth.
The message is always reassuring and always wrong. Nothing fundamental has shifted. GEO is just SEO with a rebrand. AI search is just search with a new hat. Relax. Keep publishing. Keep optimising. The fundamentals remain intact.
It is a comforting story. It is also, to be blunt about it, nonsense.
What is happening to search right now is not a cosmetic change to the results page. It is a structural change to the economics of digital discovery.
The interface is changing. The incentive structure is changing. The role that search plays in the buying journey is changing. And if you are still treating this as a technical optimisation problem, you are solving for the wrong thing entirely.
And in this article, I explain what’s happening and what you MUST do.
The economics of informational content are collapsing
Let us start with a fact that a surprising number of people in the search industry seem reluctant to fully absorb.
For two decades, the dominant logic of SEO was interception. Find the words buyers type. Build the pages that match those words. Rank higher than the competition. Capture the click.
It was a beautiful model because it was measurable, scalable, and relatively cheap once you understood the rules. An army of content writers, a content calendar, a topical cluster map, and enough patience could, for many categories, produce a steady stream of organic traffic.
That model worked because the interface required it. Ten blue links demanded ten pages. You could not answer a question that was not attached to a result. The search engine needed you.
Now it doesn’t
Large language models can synthesise thousands of sources into a single coherent answer faster than you can write the blog post. Google’s AI Overviews, Perplexity, ChatGPT, and the rest of them are not sending traffic to your informational content. They are replacing it.
The question gets answered. The user gets what they came for. And your article, the one you spent three weeks optimising for a featured snippet, sits in the supporting citations like a footnote in an essay nobody reads to the end.
If your article gets cited in an AI-generated answer, fine. But be honest about what that means. The user already has the information. They came, they received, they left.
Your reward is a small percentage of readers motivated enough to want more depth and willing to click one of several listed sources. That is not an acquisition channel. That is a rounding error dressed up as a content strategy.
The more the interface answers the question directly, the weaker the economics of pure informational SEO become. This is not speculation. It is the logical endpoint of what large language models actually do.
They consume the web’s knowledge and return it on demand. General explanatory content, the kind that fills most content strategies, is not a moat. It is raw material. And the machines are already using it.
So the question is not “how do I get cited?” The question is “how do I get chosen?”
Those are different questions with very different answers.
Network science explains what rankings cannot
Most SEO frameworks were built around a fundamentally interface-centric view of the world. A human enters keywords. A search engine interprets them. A visible ranked list returns. The game is to appear near the top of that list for the right terms.
That model made network thinking largely irrelevant. If you could rank, you could win, regardless of how well-known or well-regarded your brand actually was.
But the interface-centric model only works if the interface remains the dominant point of decision.
Once the interface shifts from a ranked list to a generated answer, a shortlist, or an agentic recommendation, the competitive logic changes completely.
You are no longer fighting to appear on page one. You are fighting to be the option that an information system, trained on the entire open web, understands and prefers.
That is a network problem.
Albert-László Barabási, whose work on network science is essential reading for anyone thinking seriously about this, identified the phenomenon he called preferential attachment. It works like this.
In any network where nodes compete for connections, links, attention, or traffic, the nodes that already have more connections are systematically more likely to acquire new ones.
The rich get richer. The well-connected get more connected. The visible get more visible. It is not a metaphor. It is a mathematical property of how real-world networks, the internet included, actually function.
Barabási called the networks that emerge from this process scale-free. They are not evenly distributed. They are dominated by a small number of highly connected hubs, surrounded by a long tail of weakly connected nodes.
And the hubs got there not simply by being better, though quality helps, but by being early, visible, and consistently referenced by others.
For brands, the implications are significant.
If the web operates as a scale-free network, then your position within it is determined not just by your own pages but by the entire web of signals pointing toward and away from you.
Mentions in trusted publications. Links from relevant domains. Reviews on authoritative platforms. Brand searches that signal demand. Creator references that extend reach. Press coverage that builds an association. Customer advocacy that confirms quality. Each one is a node. Each one reinforces your network position.
And if search engines and large language models are trained on that network, then your position within it directly shapes how well they understand you, trust you, and recommend you.
This is why the old SEO playbook starts to look thin.
Publishing generic informational content does not necessarily create more nodes in the broader network. It does not strengthen your external signal footprint. It does not change how the network understands what you do and why you matter.
Topical authority is not gained because of what you create on your own site, but in how others mention you.
And this is currently why some brands are being recommended by AI, without any optimisation, and others, highly optimised, are not.
SEOs have spent 2 decades creating signal noise to ‘trick’ Google into rewarding clients through link acquisition and content enhancements.
That is very different from building real preference, although sometimes they do cross paths. But rarely is that because of the SEO. It’s more often that the marketing work of the brand did the heavy lifting.
This is something that growth hackers got wrong. When they came in, they focused on metrics they could see, hiring SEOs to grow traffic by any means possible.
But as lessons from the growth of network products such as Uber and Facebook, as documented in Andrew Chen’s The Cold Start Solution’ show us.
Real growth comes from the network. It comes from preference.
Fitness determines who receives preferential attachment rewards
Preferential attachment does not reward everyone equally. It rewards the nodes with the highest fitness.
In network science, fitness describes the intrinsic quality that makes a node worth connecting to. Not the connections it already has, but the reasons others would choose to connect to it in the first place. A great paper gets cited more often than a mediocre one, even at equal starting visibility, because its fitness, its quality and relevance, makes it more worth referencing.
For brands, fitness is not mysterious. It is the combination of qualities that make a business genuinely worth talking about.
Clear positioning. Meaningful distinctiveness. Product quality. A point of view that is specific enough to be useful and differentiated enough to be memorable. An audience fit that is precise rather than promiscuous. The ability to create something, a report, a tool, a piece of research, a piece of advertising, a product or a service that other people actually want to reference.
This is where the SEO industry has a significant blind spot.
A great deal of SEO practice operates as though the content itself is the product. As though, if you write enough pages about the right topics in the right format, growth will follow.
But if your business lacks clear positioning, if your offer is not meaningfully differentiated, if your content is generic and your brand is invisible outside your own website, then the network has no particular reason to connect to you.
You are a node with low fitness. And low-fitness nodes do not compound.
You cannot hack your way to preferential attachment with a thin brand.
What you can do is accelerate a strong brand by making its fitness visible across the network.
Stronger PR. More useful original research. More distinctive creative. More memorable advertising. More earned mentions in more relevant places.
Each one raises your fitness score in the eyes of the people and systems doing the choosing.
That is what marketing communications has always been for.
Byron Sharp and the part SEO forgot
The marketing science here is not new. It has been sitting in the academic literature for decades, and Byron Sharp and the Ehrenberg-Bass Institute made it accessible enough that there is really no excuse for ignoring it.
Sharp’s central argument, laid out in How Brands Grow, is that brands grow primarily by increasing mental availability and physical availability. Mental availability means being more likely to come to mind when a buyer is in a relevant purchase situation. Physical availability means being easy to find, access, and buy.
These two things are not equivalent. They are complementary, and both matter.
Most of the SEO industry has, historically, focused almost entirely on physical availability. Be present in the search results. Be listed when the category is explored. Be on the digital shelf when the buyer is ready.
That is a reasonable interpretation of what search optimisation does at its best.
But Sharp’s research, replicated across categories and markets, shows that mental availability drives a disproportionate share of brand choice.
Buyers do not always search. They often choose before they search. They reach for the brand that is already salient, already trusted, already associated with the category in their memory. They use search to confirm a decision they have already largely made.
If that is true, then a brand that has invested entirely in physical availability and neglected mental availability is one bad algorithm update away from irrelevance. It was not building the preference that generates demand before the search bar is opened.
The SEO industry has underinvested in mental availability for years. Not because it did not matter. Because it was harder to measure, harder to attribute, and harder to sell to a client who wanted to see rankings move on a monthly report.
The SEO industry treated building reputation as ‘someone else’s job’.
That is about to become a very expensive oversight.
Sherwin Rosen and the winner-takes-most outcome
There is a final layer to this that most people in SEO have not thought through, and it concerns what happens once preferential attachment takes hold.
In 1981, the economist Sherwin Rosen published a paper called “The Economics of Superstars.” His argument was that in markets where talent is imperfectly substitutable and distribution costs are low, a small number of performers capture a disproportionate share of the market.
The best tennis player does not earn twice as much as the second-best. She earns ten times as much.
The reason is that the marginal difference in quality, barely perceptible to most observers, is worth paying for when your choice costs nothing extra to switch. So everyone picks the best option, and the best option wins enormously.
Rosen was writing about entertainers and sports stars. But his model describes brand markets too, particularly in categories where most buyers are not deeply engaged and where the cost of switching is low.
In those categories, once preferential attachment tips a brand into market leadership, it becomes structurally difficult to dislodge. Not because it is always the best product, but because it is the one everyone already knows, already trusts, and already defaults to. The familiarity is itself a form of value.
The accumulated network position is a competitive moat.
Google is the obvious example. It was not the first search engine. Altavista, Excite, and Lycos all preceded it.
What happened was Yahoo used Google as their search interface, giving Google massive exposure in the process. In the end, Google became so visible and so well known that it no longer needed Yahoo.
Of course, Google won because it was a better product at the right moment, and once enough people used it and linked to it and talked about it, preferential attachment compounded relentlessly. But it was that visibility to a huge audience that boosted it’s growth.
More usage. More familiarity. More default behaviour. More commercial power. Twenty-five years later, it controls roughly ninety percent of global search volume. That is what the winner-takes-most dynamic looks like fully developed.
Brands should be trying to engineer the same process for themselves.
Not at Google’s scale, obviously. But the logic holds in any category. Become the preferred option. Compound the advantage. Make it structurally harder for weaker competitors to reach the buyers who already associate you with the solution.
That is what preferential attachment actually delivers. Not just rankings. Not just traffic. An economic position that increasingly self-reinforces.
Visibility through meaning, not mechanics
The term visibility gets thrown around in marketing conversations without much precision, and it is worth being specific about what it means in this context.
Visibility is not the same as reach. A billboard on the M6 delivers reach. Whether it builds meaningful network presence depends on what happens after the impression. Does it get talked about? Does it get searched? Does it create associations that travel?
Meaningful visibility is visibility that strengthens reputation, reinforces positioning, and increases the probability that people and systems understand what you do, who you do it for, and why you are worth choosing.
In practice, building meaningful visibility means building nodes across the external network.
A node is anything that connects your brand to the wider information environment in a way that persists and propagates.
An article in a trusted industry title. A review on an authoritative platform. A data study that gets picked up and cited. A creator who becomes genuinely enthusiastic about what you do. A podcast appearance that reaches a relevant audience. A piece of original research that becomes a reference point in the category. An advertising campaign that people actually remember and talk about.
These are not all equal.
A cover story in The Economist builds a different kind of node than a five-star review on Trustpilot. But they are all part of the same system, and they all contribute to your network position in ways that generic content calendar execution does not.
Consider the contrast. A brand does a giveaway at a train station. Thousands of people see it. Momentary awareness. Possibly some positive feeling. But if the event creates no lasting signal, no coverage, no social amplification, no search uplift, no discussion, then its contribution to network position is close to zero.
Now take a brand that seeds a piece of original category research with ten relevant journalists. The research gets written up. It gets linked from three industry publications. A prominent creator makes a video responding to it. The brand gets mentioned as a source in a newsletter with a hundred thousand subscribers. Two weeks later, branded search lifts noticeably because people who saw the coverage wanted to know more.
Same budget. Completely different network impact.
That is the difference between visibility as mechanics and visibility as meaning.
Harry Potter and the compounding network
To help you grasp these concepts, we’ll stop thinking about websites for a moment and think about books.
Harry Potter did not arrive as a cultural phenomenon. It arrived as a first print run of five hundred copies, distributed to a small number of libraries and booksellers. J.K. Rowling had been rejected by twelve publishers before Bloomsbury agreed to publish it, and then Bloomsbury advised her to get a day job because there was no money in children’s books.
The book did not become the best-selling series in publishing history because someone ran a marketing campaign. It became that because a small number of early readers encountered it, loved it, told other people, and created the conditions for more people to encounter it.
One child recommended it to another. A librarian pushed it. A teacher read it aloud. A reviewer noticed it. Each connection created the conditions for more connections. The network acquired momentum. Then it compounded, catastrophically, in Rowling’s favour.
That is preferential attachment in its clearest form.
The lesson for brands is not that you need a once-in-a-generation product, though having a genuinely good one helps considerably. The lesson is that you need enough meaningful seeding in the right places to give the network something to work with. Quality plus early momentum plus consistent reputation signalling. That is the sequence.
You do not need instant domination. You need enough early nodes to start the compounding. Then you need the fitness to sustain it.
The burger bar that becomes the obvious choice
Bring it down to street level for a moment.
A local burger restaurant does not become the obvious choice in its city by publishing blog posts about brioche buns. It becomes the obvious choice by being easy to buy from and easy to talk about.
First, physical availability. It gets listed on Deliveroo and Just Eat. It has a clean, functional website. It shows up in Google Maps with a complete profile and accurate hours. It makes itself frictionless to find, order from, and experience.
Then the mental availability work begins. Customers post photos. Reviews accumulate on Google and TripAdvisor. A local food creator visits and likes what they eat. A food blogger puts it in a roundup. Someone on TikTok posts a video about the smash burger that gets thirty thousand views. The local paper runs a piece on independent restaurants worth visiting. People start recommending it to visiting friends without being asked.
More people try it because more people are already talking about it. The preference compounds. Within two years, it will be the obvious answer to “where should we eat?” for a significant slice of its city.
That is the full process in miniature. Physical availability enables the purchase. Mental availability creates the default. Network nodes amplify both. Preferential attachment does the rest.
No blog posts about brioche buns required. Although an SEO would probably still try to create one because their ‘topical map’ told them to.
Reputation signalling is the job now
If there is a single idea that should replace the old obsession with rankings, it is this: reputation signalling.
Reputation signalling is how a brand tells the market, repeatedly and credibly, what it stands for, what it is good at, and why it is worth choosing. It is how your positioning travels beyond your own website.
It is how your expertise becomes associated with a category in the minds of buyers who have not met you yet. It is how other people start saying the thing you have been trying to say about yourself.
And that last part matters enormously.
In a world where AI systems are trained on the open web, what other people say about you is more important than what you say about yourself. Your own pages tell the model what you want it to think. Third-party mentions, reviews, citations, and links tell it what the market actually thinks. Those are different things, and the models are increasingly good at distinguishing between them.
This is why PR becomes so central in the next era of search.
Not the kind of PR that chases vanity coverage for a monthly report. Strategic PR. Reputation-building PR. The kind that places your brand in the right conversations, with the right associations, in the right contexts, so that the market, human and algorithmic, understands why you exist and why you matter.
Digital PR sits in a particularly powerful position here because it does multiple jobs simultaneously.
It builds direct audience exposure.
It creates links and third-party mentions that strengthen your external signal footprint.
It lifts branded search as people who see the coverage decide they want to know more. It places your brand inside the information environment from which AI recommendations are drawn.
And it reaches actual human beings, which remains, despite what some digital absolutists would have you believe, rather important.
The same logic applies to content marketing, with one significant qualification. Content that builds reputation, that gives the market something original, useful, and genuinely worth referencing, compounds.
Content that fills a topical cluster spreadsheet does not.
Three white papers that the whole category reads and cites can do more for your network position than three hundred SEO-optimised blog posts that nobody shares and nobody remembers. The distribution of value in content is not linear. It follows the same power law that governs the rest of the network.
This isn’t a process just for large brands. The burger bar that posts a video of their food each week and shares it on social.
The local accountant who creates a report each month with finance tips for small local businesses in their area.
The engineering firm that interviews staff at work and asks them to explain what they do and why it matters for the end product, placing the content in their newsletter, on YouTube, on their website and on LinkedIn.
This content doesn’t have to go viral. It’s not supposed to. JK Rowling didn’t go viral at first.
This content is to build signals to both humans and machines, one signal at a time.
What this means for people working in SEO
None of this argues that search expertise becomes irrelevant.
Someone still needs to understand the technical architecture of the website.
Someone still needs to manage the product and service pages that capture in-market demand.
Someone still needs to ensure the digital experience is functional, fast, and clear. Someone still needs to interpret what is happening in the search landscape and translate it into commercial action.
As agentic commerce develops, and as AI systems become more active participants in the buying journey, that kind of analytical and commercial fluency will become more valuable, not less.
But the narrow discipline, the one that treats traffic as the mission and rankings as the strategy, is under serious structural pressure.
If your entire model depends on ranking informational content, tweaking metadata, and producing monthly reports that show traffic going up, you are exposed.
The informational content layer is being absorbed by AI systems. The traffic-as-mission model has always been a proxy measurement rather than a real one.
And the report showing ten thousand page views to an article nobody converted from is not a defence when a client starts asking harder questions.
The growth roles of the next decade will belong to people who can hold search, content, PR, brand, analytics, and distribution in their heads simultaneously and understand how they interact.
The winners will think like marketers. They will understand brand building, not just optimisation. They will be comfortable with brand investment that does not convert in a ninety-day attribution window.
The agencies that adapt will restructure accordingly. The disciplines are collapsing into each other. Social, PR, content, brand, search, and analytics are becoming facets of a single question: how do you build a brand’s reputation and preference across the network?
That is a marketing question. It always was.
What to actually do
Make your commercial pages excellent. Product pages, service pages, solution pages, category pages. These are the pages that capture in-market demand when it exists. They should be clear, credible, and optimised for the buyer who is already close to a decision.
Invest in external signal quality. Not links for the sake of links, but earned coverage, credible reviews, relevant partnerships, creator associations, and industry recognition that collectively tell the network something real about your brand’s position and quality.
Create content that earns its place. Original research. Distinctive points of view. Useful tools. Data studies. Practical resources. Pieces that a buyer or journalist would actually want to reference. Not content that fills a template.
Monitor what matters. Branded search volume is a proxy for reputation. Quality external links and mentions signal network position. Category association in relevant publications signals positioning progress.
Revenue, obviously, is the point. Traffic can go back to being a useful diagnostic metric rather than an end goal.
Above all, ask a different question. Not “can we rank for this?” but “does this build preference?”
If the answer to the second question is no, reconsider whether the activity belongs in the budget at all.
Build visibility. Build reputation. Build preference. Let the network do the rest.
Brands that understand this now will find the next era of search more hospitable than those still waiting to be told the fundamentals haven’t changed.
They have.
Andrew Holland
Originally published on LinkedIn on 31 March 2026.