Search marketers are having a miserable time of it.
Traffic is slipping. Informational content is being chewed up. AI Overviews are stripping clicks away from the old blog-first model. ChatGPT, Gemini, Claude and the rest are changing how people discover businesses, compare options and enter buying journeys.
And the response from much of the SEO industry has been exactly what you would expect from an industry that has spent too long staring at its own spreadsheets: panic, simplification and tactical bollocks.
Should we chase citations? Should we make more listicles? Should we publish ten more pages this month? Should we build a dashboard for “AI visibility” and pretend that means we understand anything?
No.
Because the biggest issue in search right now is that the industry is looking for answers to its current problems inside the same worldview that created them. And it will not find them there.
Search is not mainly a keyword problem anymore. It is a network problem. And if you keep trying to solve a network problem with old SEO thinking, you are going to keep getting seduced by the wrong things.
The circle of knowledge problem
One of the oddest things about SEO is how often the industry admits it doesn’t know what’s going on while simultaneously behaving as though it already has the answer.
That is not wisdom. That is a circle of knowledge. You don’t know what you don’t know.
Most SEOs are not marketers. They did not come up through strategy, brand, market orientation or positioning. They fell into search. They inherited a worldview from the agencies they worked at, the blogs they read, the conferences they attended and the people they copied. And then that worldview got repeated until repetition itself started looking like truth.
Popular things said in SEO become more popular because they are already popular. Someone says something, it gets shared, screenshotted, paraphrased, copied into pitch decks, turned into a framework, handed to juniors, and repeated on LinkedIn by someone who needs to make sales.
That is memetics. And it matters because it mirrors a lot of what happens in search itself. Visible things become more visible because they are already visible.
Popular ideas get copied because they are already popular. The rich get richer. Not because they are always right, but because they are already there.
Traffic was the classic example.
Traffic started as a measure. Then it became a target. And the second that happened, the industry found endless ways to produce more of it.
How traffic became the addiction
Once traffic became the thing, content production became the machine.
Most of the growth era in SEO was built around informational publishing. Blog content. Advice content. Top-funnel guides. Endless articles for queries adjacent to the business rather than central to it. It was easy to justify because traffic was easy to report, easy to graph and easy to sell. The client could see it. The agency could point at it. Everybody could pretend the game was working.
But go into the analytics of almost any large SaaS site and you will usually see the same grim pattern. A tiny number of pages drives most of the traffic. Sometimes less than 0.5% of the site pulls in 80% or more of organic visits. The rest of the site, often thousands of pages, contributes a miserable minority.
Now do the maths.
If each decent article costs £1,000 once you factor in writing, editing, upload, design and management time, then a site with 6,000 articles may have spent millions creating a giant archive of content that contributes bugger all commercially.
That is not strategy. That is accumulation mistaken for progress.
And yet, to be fair, not all of it was stupid.
Not all informational content was fluff
One of the dumbest habits in digital marketing is replacing one simplistic dogma with another. First, “content is king”. Then, when that starts to wobble, “blog content is dead”. Both are lazy.
Some informational or inspirational content did a very real commercial job.
Take a bespoke kitchen company ranking for “kitchen ideas”. Or a bathroom company ranking for “bathroom ideas”. That traffic is not automatically vanity traffic. If a buyer is entering the category, looking for inspiration, and repeatedly seeing your products at the exact moment they are mentally building a shortlist, that is smart marketing. Your product is being placed in the consumer’s mind while they are coming into market.
That is not fluff. That is upstream demand shaping.
Likewise, some content was built to attract links rather than immediate conversions. Power pages, tools, original resources, assets designed to earn links and strengthen the commercial pages underneath them. That also made sense. Much of classic SEO that actually worked was not really about traffic at all. It was about building pages that could acquire links and then pass strength into the pages that made money.
So this is not some stupid argument that all content was a waste of time. It is a more precise point than that.
The problem is that the traffic era trained SEO to obsess over visible outputs and then confuse those outputs with the real mechanism underneath.
And that bad habit is now being dragged straight into AI search.
Why the citation obsession is so dim
Take the example of an SEO agency in Manchester.
There are plenty of them. Let’s say fifty. Most are broadly similar. They all want to rank for the same core commercial terms: “Manchester SEO agency”, “SEO Manchester”, “SEO services Manchester”, and so on.
In the old world, page one mattered because users needed a shortcut. Nobody had time to evaluate fifty agencies properly, so ranking became a proxy for quality.
If you ranked, you looked credible. If you did not, you were ignored.
Now, imagine the user is not typing into Google. They are asking an LLM:
“Find me the best SEO agencies in Manchester.”
That sounds similar, but it is not the same task at all.
You are no longer asking for pages that match a phrase. You are asking for a judgement. Which agencies are best? Which are most likely to solve the problem? Which should be shortlisted?
And that is where most of the GEO conversation collapses into childish nonsense.
The model cannot truly know who is best. There is no objective “best” sitting there waiting to be collected. It has to infer. It has to reduce uncertainty. It has to draw on what it already knows, what it can retrieve, what it can corroborate, and then produce the most plausible answer it can.
So when SEOs look at the output and see that an LLM cited a “best SEO agencies in Manchester” article, they leap to the wrong conclusion. They assume the citation caused the recommendation.
It probably did not. Or at least not on its own.
The citation is evidence of part of the process. It is one visible layer. It is not the end result, and it is not automatically the determining factor. The model may already know the names of agencies from its parametric knowledge. It may also use retrieval-augmented generation and live search to find supporting information, recent references and grounding. It may fan out into multiple searches and compare different kinds of evidence. The citation is just what you can see.
That is the crucial mistake the industry is making. It is conflating the surface artefact with the underlying mechanism.
Because SEO has always loved doing that. See a thing. Mistake it for the cause. Industrialise it. Ruin the web with it.
Known risk, unknown risk
When I was in the police, we did not think in terms of risk and no risk. We thought in terms of known risk and unknown risk.
That is a much better way to think about AI search than the false certainty currently being sold in the market.
We know some things. We know these systems have internal knowledge. We know they can retrieve external information. We know they ground themselves. We know there are methods for combining ranked outputs and retrieved evidence. We know different systems work in different ways.
But we do not know the actual recipe.
We do not know the exact prompt the user typed or spoke. We do not know how it was decomposed. We do not know the fan-out queries. We do not know the weighting between parametric knowledge and retrieved evidence. We do not know what was ignored. We do not know what was privileged. We do not know why one source made it into the output while another merely informed the process in the background.
And because the SEO industry is full of convergent thinkers who need things to be black and white, it keeps trying to force certainty onto a situation that is still, in truth, partly unknowable.
That is why the answer cannot be “go get more citations”.
If you cannot see the prompt, cannot control the retrieval path and cannot know the exact inference chain, then optimising for the visible citation is fragile at best and idiotic at worst.
You have to go one layer down.
Stop optimising for the interface. Start optimising for the network.
This is the real reversal.
For years, search strategy was built around the interface. Google was the screen. Keywords were the unit. Rankings were the prize. We optimised for the end display.
That is no longer enough.
Now you have to optimise for the network beneath the interface. The web. The wider information substrate. The places where your business can be encountered, understood, corroborated and connected to the problems it solves.
And once you accept that, the whole strategy changes.
A user does not really have a keyword. They have a problem.
The more specific they are, the more constrained the prompt becomes. “SEO agency Manchester” is broad. But “I’m an accountant in Manchester with a WordPress website and I need SEO support from someone who understands my sector” is much tighter.
Now think about what an LLM may have to do with that.
It may search for WordPress specialists. It may look for SEO consultants, not just agencies. It may weigh sector experience with accountants. It may look for local providers. It may compare service pages, reviews, directories, case studies and adjacent language. It may bring back a mixed list: one bigger agency, one specialist consultant, one niche WordPress firm, one freelancer with accounting experience.
You do not see that logic. But you do know yourself.
You know what problem you solve. You know who you solve it for. You know how you solve it better. You know what your proof is. And that is what your website, your service pages and your external visibility should be built around.
Not the phrase. The position.
That is what positioning is. What problem do we solve? For whom? Why us? What evidence have we got? What is our recipe?
And here is the funny part: this is not some bizarre new AI discipline. It is simply proper marketing arriving late to search.
The consultant story
Imagine an SEO consultant leaving a big agency in Manchester.
They have worked there for years. They know WordPress cold. They have seen hundreds of client sites. They know how to improve site speed, fix technical issues, lift conversions and help local firms show up properly in the map pack and on high-intent commercial searches. They have done it before. They are competent, experienced and more cost-effective than the bigger agencies they have just left.
So they decide to go out on their own.
What do they do?
They do not sit in a room and dream up 500 informational keywords about SEO.
They build a business.
They create a brand. They position it. Maybe they decide to own “WordPress SEO for small businesses in Manchester”. Maybe they make the decision to work only with WordPress because they know it inside out and do not want to waste time on platforms they are weaker on. Maybe they decide to focus on speed, conversions and local search visibility. That becomes the core of the offer.
Then they put it into the world.
They build a site that says exactly what they do. Not generic agency wank, but proper positioning. They help Manchester small businesses on WordPress. They improve website performance. They strengthen local visibility. They help firms turn traffic into leads. They say who they serve and how.
Then they start grafting.
They set up social profiles. They go to local events. They join the Chamber of Commerce. They offer a workshop for local business owners on how to make WordPress websites faster. They write a practical guide for small businesses on what actually helps a site convert. They get listed in directories local firms use. They ask clients for reviews. They publish case studies. They partner with a PPC freelancer and a local web designer. They refer work. They get referred back.
Then they start trying to be found where their audience actually is.
A local business site lets them guest-write a piece on how speed improvements can lift enquiries from a WordPress site. A Manchester business newsletter mentions their workshop. A local publication includes them in a piece on digital support for small firms. They get asked to speak at a breakfast event. They run a webinar. They create a downloadable checklist that gets shared around local business groups.
This is just normal marketing. But it is also exactly how a business becomes legible inside a network.
The links matter, yes, but not only because they pass authority in the old SEO sense. They matter because they are nodes. They create pathways. They create encounter points. They make the consultant visible in contexts that match the problems they solve.
Over time, something starts to happen. The consultant becomes known as the WordPress SEO person in Manchester. They get referred more easily. They get mentioned more often. People start to think of them without needing to search. Their success starts to make future success easier.
That is preferential attachment.
What Albert-László Barabási helps explain
This is where network science is actually useful, not just a clever phrase dragged into a pitch deck.
Albert-László Barabási’s work on networks helps explain why certain nodes grow faster than others. Some gain links, attention and connections more easily. That is the rich-get-richer effect of preferential attachment. But in real markets it is not only previous success that matters. It is also fitness.
A business has performance characteristics that make it more likely to out-compete others in a network. It solves a problem well. It does so for a specific group. It has proof. It has relevance. It has differentiating traits. It has visibility.
That is why some businesses naturally attract more nodes than others. They are fitter. And if they are also visible, they start to compound.
Fitness without visibility does not travel. Visibility without fitness does not last. Put them together and you get growth.
That is why the consultant can start from scratch and grow. Not because they gamed Google better, but because they built a position, made it visible, created encounters and slowly built enough presence that the network started to work for them.
Then they get ambitious.
They no longer want only Manchester small businesses. They want bigger clients. Maybe they have just landed a larger law firm and done a brilliant job. Now they have a choice. They can change the positioning or extend it. Maybe they decide to build a new arm of the consultancy focused on law firms.
What do they do then?
Exactly the same thing, but in a new network.
They create pages that speak to legal-sector problems. They develop legal case studies. They attend legal marketing conferences. They get into directories law firms actually use. They pitch guest articles to legal publications. They run webinars for law firms. They start showing up where legal buyers and legal marketers can encounter them. They build the co-occurrences, mentions and associations that make them visible in that vertical.
And after enough of that, the same compounding begins again.
They become one of the names that comes up in legal SEO conversations. Then one of the names buyers shortlist. Then one of the names machines repeatedly encounter when users have prompts related to law firm SEO support. The richer network around them grows because the business has become more visible and more fit for that problem.
This is business development. This is PR. This is communications. This is content marketing. This is local SEO thinking applied properly. And, increasingly, this is what AI search rewards.
What GEO actually is
Generative engine optimisation is not just SEO with some AI glitter thrown over the top.
Yes, it still has on-page and off-page elements. But the nuance is different.
On-page, you are not merely matching a keyword. You are isolating and articulating your fitness signals. You are making your business machine-legible around the problems it solves.
That means who you help, where you help, how you help, why you are different, what evidence you have and what specialism matters. If you sell young driver insurance and the hidden prompt is from an 18-year-old who has just passed their test, it might matter that your page explicitly says you help drivers aged 17 to 24.
Not because you need a ridiculous separate page for every age, but because specificity matters. If you help certain cities, say so. If you specialise in WordPress, say so. If you work with accountants, say so. If you only work with Shopify DTC brands, say so.
And that is where the contradiction between the old and the new starts to show up.
Old SEO would often tell you to broaden, add more topic coverage, create more supporting content, build more topical relevance. GEO forces you to tighten. Clarify. Position. State what you are. State who you serve. State how you solve it. If you bury that under five paragraphs of subject waffle, you are hiding the very thing the machine is trying to isolate.
A law firm page does not need five vague paragraphs about “the legal landscape” before it gets to what the solicitor actually does. A service page does not need a small essay on the history of insurance before it states who it is for. A business page should not ramble around the subject and bury the offer under topic sludge. That is old-web copywriting and weak SEO thinking colliding with the new reality.
GEO demands cleaner positioning.
Off-page, the same principle applies. You are taking those fitness signals and pushing them into the network. PR, reviews, directories, partnerships, guest articles, tools, communities, events, social distribution, founder presence. The goal is not just “more links”. The goal is more meaningful visibility around what the brand actually does well.
That is why GEO sits much closer to PR and content marketing than most people want to admit. It is marketing communications. It is the work of making a business more famous for the problem it solves.
That is why I call it fame engineering.
Why spammy tactics sometimes still work
One final nuance, because the internet is messy and stupid.
Yes, some spammy tactics still seem to work. Thin list-post ecosystems. Over-presence in weak directories. Social blasts at scale. Why?
Because when a market is shallow and badly differentiated, small signals can tip the machine.
If fifty SEO agencies in Manchester all look roughly the same and none has built a meaningfully deeper network than the rest, then being present in ten extra listicles may make one easier to recommend. That does not mean listicles are the strategy. It means the market is under-signalled.
When there is no clearly more famous, more visible, more coherently positioned option, smaller tactical artefacts can nudge the system.
But that is not a moat. It is a temporary edge in a weak field.
SEO is not dead. But it is no longer enough on its own.
This is where the usual idiots get overexcited and start announcing the death of SEO.
It is not dead.
Keywords still matter. Site architecture still matters. Internal linking still matters. Technical competence still matters. Commercial pages still need to be optimised. Organic search still drives value. Good SEO is not disappearing.
But GEO is not the same thing, and pretending it is merely “SEO for AI” is too simplistic.
On-page SEO and on-page GEO are now related but different disciplines. One helps you align with search demand and structure pages well for traditional discovery. The other helps you clarify position, fitness and usefulness so machines can isolate what you are and why you matter.
Off-page SEO and off-page GEO are related but different too. One historically thought in terms of links, authority and rankings. The other has to think in terms of meaningful visibility, co-occurrence, reputation, contextual mentions, recommendations and presence in the network.
So these two should not be at war. They should not be treated as rival religions. They are additions to each other. Separate ways of thinking. Separate ways of doing. Shared ground, but different emphasis.
Traditional SEO still asks, “What keywords should this page rank for?” GEO asks, “What exactly is this business fitted to solve, and where must that be visible so it becomes recommendable?”
That is a much better tension to work with than the current nonsense.
The real job now
The real job of the search marketer is no longer to sit in the corner fiddling with keywords and reporting traffic as though that is the entire commercial picture.
The real job now is harder, broader and more useful.
It is to identify a business’s fitness signals. To isolate what it actually does well. To make that positioning brutally clear on the page. To strip away waffle that buries the offer. To build content that reflects real problems, real buyers and real use cases. To increase meaningful visibility for that business across the network in the places where buyers and machines encounter evidence. To connect copywriting, content marketing, PR and SEO into something that actually resembles business communications.
That is why this is difficult for agencies.
It does not fit neatly into the old SEO retainer. It is harder to package than “four blog posts and ten links a month”. It does not always sit inside one channel. It spills into PR, social, reviews, events, partnerships, case studies, site messaging and founder visibility.
It asks tougher questions. What are we famous for? What should we be famous for? What problem do we solve better than others? Where does the market need to encounter that?
That is why many SEO agencies will struggle with it. Not because the work is impossible, but because it is less tidy. It is more commercial. It is more strategic. It looks less like factory SEO and more like real marketing.
And that, frankly, is the point.
Search can grow as a channel. SEO can remain powerful. GEO can add to it. But only if the industry stops obsessing over the wrong things.
Not citations for their own sake. Not traffic for its own sake. Not topical sprawl for its own sake. Not flat metrics and vanity dashboards.
The real job is visibility.
Not empty visibility, but meaningful visibility.
The real job is isolating the fitness of a business, then increasing the meaningful visibility of that business online within a network until it becomes easier for humans and machines to encounter, understand and recommend.
That is the work.
That is the future.
And it looks a lot more like copywriting, content marketing and PR than most search people are comfortable admitting.
Andrew Holland
Originally published on LinkedIn on 27 March 2026.