There is a peculiar habit in marketing. When something stops working as well as it used to, we do not question the underlying logic. We invent new metrics.
That is exactly what is happening in search right now.
SEO is under pressure because informational traffic is falling.
GEO has appeared as the supposed solution, complete with shiny new dashboards, new acronyms, and a new promise to calm nervous CMOs.
The problem is that neither discipline is being judged by the right standard. Worse, both are leaning on metrics that look commercial but explain very little about growth.
This is not a search problem. Instead, it’s a huge measurement problem. One that will cause a lot of headaches.
Marketing has been here before. For decades, marketers have clung to ROI as the ultimate proof of competence, even while some of the smartest people in the discipline have warned that it is one of the most misunderstood and misused metrics in business. Now that confusion has bled into SEO, and GEO is doubling down on it.
The result is predictable. Budgets get justified on the wrong numbers. Agencies sell activity instead of outcomes. Boards get reassured while growth quietly stalls.
If this article has one goal, it is not to kill SEO or dismiss GEO. It is to make them economically intelligible again.
That requires going back to basics. Before we talk about traffic, placement, influence, or AI recommendations, we need to clear up the single concept that sits underneath all of this.
ROI.
ROI is a ratio, not a strategy
Return on investment is not a growth metric. It is not a strategy. It is not even, strictly speaking, a measure of effectiveness.
ROI is an accounting ratio. Net profit divided by cost.
That distinction matters more than most marketers are comfortable admitting.
Effectiveness is about scale. How much profit or revenue did this activity generate. Efficiency is about how cheaply it did so. ROI only tells you the second part. It says nothing about whether the activity was big enough to matter.
This is why ROI has such a long and uncomfortable history in marketing. Time and again, data shows the same paradox. As marketers focus harder on efficiency, overall profit declines. Campaigns get smaller. Targeting gets narrower. Ambition evaporates. ROI goes up while the business goes backwards.
The reason is simple. The fastest way to improve ROI is not to grow the numerator. It is to shrink the denominator. Spend less money. Do smaller things. Chase marginal gains. This looks impressive in a spreadsheet and disastrous in a P&L.
Marketing effectiveness research has been remarkably consistent on this point. Budget and scale explain far more variation in profit than ROI ever will. Yet marketers continue to behave as if efficiency is the primary driver of success, largely because they believe this is what boards want to hear.
It is not.
Boards want growth that is defensible and sustainable. They want profit. They want customers. They want evidence that demand is expanding, not just that costs are under control. ROI is useful as a sense check. It is dangerous as a decision rule.
This is why some of the sharpest thinkers in marketing have been so blunt about it. ROI is not wrong. It is just routinely asked to do a job it was never designed to do.
Once you accept that, a lot of things start to make sense. Including why entire channels get misjudged.
Search is a perfect example.
SEO did not become confused because practitioners were stupid. It became confused because ROI thinking rewards the appearance of efficiency over the reality of contribution. When you apply that logic to search, the industry naturally gravitates toward whatever looks cheapest and easiest to show on a chart.
That is how traffic became the star of the show.
And it is how SEO learned to sell the wrong thing.
How SEO learned to sell the wrong thing
Once you understand how ROI rewards efficiency over scale, the evolution of SEO starts to look inevitable rather than accidental.
In the early days of the internet, SEO genuinely worked the way people still talk about it today. There was very little content online. Competition was limited. Paid search was cheap. If you published something useful and vaguely optimised it, you ranked. Traffic flowed. Businesses grew.
Under those conditions, traffic was a reasonable proxy for value. More visitors usually meant more customers, or at least more people who might become customers later. SEO earned its reputation as a high return channel because, at that moment in time, it was.
This is the period that produced the foundational myths of the industry. Big guides. Educational blogs. Long form content that genuinely helped people understand a new discipline. When Moz published its early SEO guides, when Neil Patel built Quick Sprout, when Backlinko emerged with deep dives into backlinks and ranking factors, the model made sense.
People reading that content were in the market. They wanted better search performance. They were deciding whether to learn, buy tools, or hire help. Content worked because it aligned directly with commercial intent.
The problem is that this success became the template long after the conditions that made it work had disappeared.
As the web matured, content exploded. Every category filled up. Every obvious keyword was targeted. Competition intensified. At the same time, SEO agencies discovered that content was not just effective. It was scalable, repeatable, and easy to package.
Two blog posts a month. Four blog posts a month. Twelve blog posts a quarter. Graphs showing traffic growth. Simple narratives that reassured clients and boards alike.
Content became the deliverable because content looked efficient.
Links were harder. Technical SEO was riskier. Buyer intent keywords were competitive and volatile. Informational content was safer.
It was cheaper to produce in volume and easier to defend in reports.
This is how SEO drifted. Not because content stopped having value, but because traffic became easier to sell than contribution.
The industry slowly stopped asking whether the traffic mattered and focused instead on whether the line on the chart went up.
Which brings us to the point where things broke.
Goodhart’s Law and the collapse of traffic as a target
There is a simple rule that explains what happened next. When a measure becomes a target, it stops being a good measure.
SEO did not break traffic. It optimised it to death.
Once traffic became the goal, everything bent toward maximising it. Keywords were chosen for volume rather than intent. Content was written to rank rather than persuade. Success was framed as sessions, impressions, and growth curves, detached from revenue or customers.
This is where concepts like topical authority flourished. If you publish enough content around a theme, the theory goes, search engines will reward you. And in many cases, they did. At least for a while.
The problem was not that this traffic was useless. The problem was that it was increasingly untethered from commercial reality. Informational visitors arrived, consumed content, and left. They were counted, celebrated, and reported. They rarely bought anything.
Traffic became a vanity metric not because it was inherently meaningless, but because it was abundant and easy to game. The signal got drowned in noise.
Then something interesting happened.
AI arrived, and informational clicks started to disappear.
Search engines began answering questions directly. Summaries replaced blue links. The endless top of funnel content that SEO had relied on for years stopped delivering the same volume of visits.
At first, this looked like an existential crisis. In reality, it was a correction.
Because when traffic is no longer the thing everyone can inflate, it starts to become meaningful again.
Not all traffic. Specific traffic.
Traffic to places where money changes hands.
The irony is hard to miss. The industry spent a decade turning traffic into a target and stripping it of meaning. Now that traffic is getting scarcer, it regains its value, but only where intent is real.
That realisation forces a difficult question. If most SEO traffic never mattered commercially, which part ever did?
To answer that, we have to separate SEO traffic into two very different types. One that always looked impressive and one that quietly paid the bills.
The only SEO traffic that ever really mattered
If traffic is starting to mean something again, the obvious question is which traffic is worth caring about.
The answer is unglamorous and inconvenient. Buyer intent search.
Buyer intent search is where people are actively trying to solve a problem with money. They are looking for a product, a service, a provider, or a comparison that helps them decide who to pay. This is the territory that Google Ads has dominated for two decades for a simple reason. It works.
If paid search did not generate revenue, Google would not be the business it is today. The existence of ads on a search results page is the clearest signal you will ever get that commercial demand exists. People bid for those positions because customers convert there.
Organic search sits directly underneath that paid shelf space. Ranking organically for those same queries is not content marketing. It is placement.
This is the fundamental mistake that has warped how SEO is discussed.
SEO is not primarily about educating the masses or building a library of thought leadership. At its most commercially valuable, SEO is about being present where people are already trying to buy.
That does not mean informational content has no role. It means its role has been massively overstated. Informational SEO works best when the subject matter is tightly aligned with what you sell and when the audience reading it is plausibly in the market. This is why SEO tools, marketing education platforms, and early SaaS businesses did so well with content. The education was an extension of the product.
In most categories, that alignment breaks down quickly. Writing about a broad topic does not mean the reader understands or values your specific offering. Awareness without relevance is cheap and often useless.
Buyer intent traffic behaves differently. Volumes are lower. Competition is fiercer. Rankings are harder to win and easier to lose. But the economics are clear. If someone lands on a page where you sell something and converts, the value is immediately legible.
This is where traffic becomes meaningful again. Not because it is large, but because it is expensive to replace.
If you lose that traffic, you have to buy it. Usually from Google. At a known cost per click.
This is the only sensible way to think about the value of SEO traffic. What would it cost you to acquire the same visitors through paid channels, and what do those visitors generate over their lifetime?
Once you frame it that way, SEO stops being mystical. It becomes an investment decision.
Which leads directly to the next uncomfortable question. What does SEO actually cost, and is it worth it?
The real economics of an SEO retainer
Most SEO retainers look similar once you strip away the language.
There is some strategy. There is keyword research. There is on-page optimisation. There is link building, often outsourced. And there is content. Lots of content.
Content dominates because it is the easiest thing to produce, the easiest thing to invoice, and the easiest thing to justify in a report. It also happens to be one of the highest margin components of an SEO agency’s work.
A typical mid-range SEO retainer might cost between £20,000 and £25,000 per year. In return, a client gets a steady flow of articles, some optimisation work, and a promise of growing organic visibility.
The implicit deal is simple. Spend this money, and you will get more traffic.
The harder question is whether that traffic is worth the spend.
If an SEO programme delivers a few hundred or even a few thousand additional visitors a year, the commercial impact depends entirely on who those visitors are and what they do. If they are loosely interested readers who never buy, the ROI conversation collapses quickly.
This is where ROI thinking re-enters the story, and why it causes so much confusion.
The right way to judge SEO is not to ask whether traffic increased. It is to ask how many customers were acquired and what those customers are worth over time.
This is not perfect attribution. It does not need to be. It is basic commercial reasoning.
The problem is that many SEO programmes never get close to this level of clarity. They trade in abstractions. Compounding value. Authority. Visibility. They rely on the idea that content will eventually pay off without being specific about how or when.
In a world where informational traffic is shrinking, that ambiguity is no longer sustainable.
Once you reduce SEO to placement and customer acquisition, you can decide whether it deserves budget. You can also see why so many SEO retainers struggle to survive scrutiny.
And this is precisely why GEO has emerged with such force. When one story stops working, another one is invented.
Unfortunately, the new story is even harder to justify.
GEO and the rise of the grifter metric
When SEO stopped being easy to justify, the industry did not pause to rethink its foundations. It rebranded.
GEO, AEO, AIO, generative engine optimisation. Pick your acronym. The promise is always the same. AI has changed search, therefore you need a new service. That service will give you visibility inside large language models. Your content will be cited. Your content will be mentioned. Your content will appear in answers.
And you will have no idea whether any of it is worth paying for.
This is where GEO develops its grifter problem. And yes, despite all their noise, some of the most white-hat, industry-celebrated SEO gurus are now becoming GEO/AI SEO grifters by trying to sell the same deliverables under a different name.
Or worse, use the same name, “it’s just SEO”. It’s a giant grifters club of people who want to sell and optimise articles. All while pulling the Spiderman meme of ‘calling everyone else a GEO grifter’.
The dominant metrics being sold today are things like AI visibility, citation frequency, prompt presence, or model inclusion. They sound impressive. They look modern. They are also commercially empty technowaffle.
A citation with no click has no direct economic value. A mention without recall builds little brand equity. Having a few lines of content present in a synthetic answer does not tell you whether anyone was persuaded, influenced, or converted.
Worse, these metrics cannot be audited in any meaningful way. You cannot see the full range of prompts people use. You cannot observe the decision process inside a model. You cannot isolate cause and effect. What you get instead are proxies layered on top of assumptions.
This is fertile ground for selling confidence without accountability.
It is not that AI recommendations do not matter. They matter enormously. It is that the industry has chosen to measure the wrong thing because the right thing is harder to sell.
Influence does not show up neatly in dashboards. Brand preference does not spike overnight. Recommendation effects show up indirectly, through changes in demand, search behaviour, and conversion patterns.
Rather than grapple with that complexity, many GEO offerings retreat into surface-level signals. They confuse being seen with being chosen.
This is the same mistake SEO made with traffic, only faster.
The uncomfortable truth is that most of what is being sold as GEO today is not optimisation at all. It is content production dressed up as inevitability. Feed the models. Publish more. Get cited. Trust the process.
That is not strategy.
To understand what GEO is actually good for, you have to abandon the idea that it is a technical discipline. It is not. GEO is influence.
And influence works very differently.
GEO is influence, not optimisation
The moment you stop thinking of GEO as optimisation, it starts to make sense.
Large language models do not rank pages the way search engines do. They synthesise information from sources they deem credible, popular, and representative. They echo what they see most often, what is reinforced across channels, and what aligns with established narratives.
The return on GEO does not come from citations. It comes from recommendation.
Being named as a good option when someone is deciding what to buy.
Being surfaced as a default choice.
Being familiar enough that both humans and machines reach for you first.
This aligns almost perfectly with what brand marketers have known for decades. Fame matters. Fluency matters. Positive feeling matters.
If a brand is widely talked about, consistently associated with specific strengths, and easy to recognise, it gets recommended more often. By people and by systems trained on people’s behaviour.
Seen through this lens, effective GEO activity looks nothing like traditional SEO deliverables.
It looks like PR. It looks like reviews. It looks like list placements, creator content, trade coverage, community discussion, and third party endorsement. It looks like being present in the places models draw from because those are the places humans trust.
A hotel does not improve its chances of being recommended by an AI system by chunking its homepage. It improves its chances by being reviewed, discussed, filmed, listed, and remembered.
A SaaS company does not get recommended because it has schema. It gets recommended because it appears repeatedly in credible discussions about the problem it solves.
The optimisation happens upstream. In what you choose to promote, who you persuade to talk about you, and where those conversations take place.
This is why GEO is exciting when done properly. It reconnects search with real marketing. It rewards businesses that understand their strengths and are willing to invest in being known for them.
It is also why GEO is uncomfortable for many agencies. Influence is harder to package than content. It involves negotiation, partnerships, media, and reputation. It does not scale neatly. It does not sit cleanly inside a fixed monthly deliverable.
But that discomfort is the point.
Once you accept that GEO is influence, the next question becomes practical rather than theoretical. What does this actually look like in the real world, and what kinds of activity move the needle?
That is where the conversation finally gets useful.
What effective GEO actually looks like in practice
Influence sounds abstract until you see it operating in the wild.
Take a hotel in Jamaica. When someone asks an AI system for the best hotels to stay in Jamaica, the model does not invent an answer. It draws from what already exists. Reviews, travel blogs, YouTube videos, listicles, forum discussions, booking platforms, and media coverage.
If a hotel consistently appears across those sources, in a positive light, associated with specific strengths such as weddings, luxury, family friendliness, or value, it becomes easy to recommend. Not because it was optimised, but because it is well represented.
This is where GEO becomes practical.
The work is not about feeding content to models. It is about deciding what you want to be known for, and then ensuring that story appears repeatedly in the places that matter.
That might mean partnering with travel creators and giving them a real experience worth talking about. It might mean securing placements in credible travel publications. It might mean actively encouraging and responding to reviews on platforms that models frequently reference. It might mean sponsoring content, funding research, or creating assets that others want to cite.
The same logic applies in other categories.
A restaurant in London does not win GEO by publishing articles about steak. It wins by being reviewed, discussed, filmed, and recommended as a place people genuinely enjoy.
A SaaS company does not win GEO by generating endless blog posts. It wins by being mentioned in industry conversations, comparison pieces, case studies, and peer recommendations.
What matters is not volume of content, but concentration of meaning.
The most effective GEO programmes are selective. They focus on specific buying situations, specific audiences, and specific narratives. They aim to flood the ecosystem with consistent signals, rather than scatter weak ones everywhere.
Crucially, this work has value even if AI did not exist. Humans see it. Humans are influenced by it.
AI simply amplifies what is already present.
This is why GEO done properly feels like a return to marketing basics rather than a technical breakthrough. It is promotion with intent.
And it is precisely why it does not fit neatly into the old SEO agency model.
Why SEO and GEO pricing will break
Once you accept that SEO is placement and GEO is influence, the economics of most agency retainers stop making sense.
Traditional SEO pricing is built around production. Blog posts. Optimisation tasks. Links. Monthly outputs that can be itemised and repeated. This model worked when content volume correlated with traffic and traffic could be defended as value.
GEO breaks that logic completely.
Influence cannot be templated. It involves strategy, prioritisation, negotiation, and judgement. The execution often happens outside the agency, through creators, publishers, platforms, and communities. The costs are variable and frequently paid directly to third parties.
This shifts where the value sits.
Strategy becomes the premium. Knowing which battles to fight, which narratives to push, and which channels to invest in becomes more important than producing assets. Delivery becomes cheaper and more distributed. AI reduces production costs. Media spend and partnerships become more transparent.
For agencies, this is uncomfortable. It means fewer easy retainers and more upfront thinking. It means clients questioning what they are paying for. It means margins moving away from content and toward brains.
For clients, it is also uncomfortable. Paying for thinking feels harder than paying for outputs. Negotiating with influencers, publishers, and platforms feels messy compared to approving a monthly content plan.
But this is where effectiveness lives.
In the long run, agencies will split. Some will double down on low cost production and compete on volume. Others will specialise deeply, build networks, understand niches, and charge for strategy and access.
GEO accelerates that divide because it rewards those who understand marketing, not just mechanics.
This is why so much of the current GEO conversation feels hollow. Selling AI visibility is easier than selling influence. Dashboards are easier than decisions. But ease has never been a reliable indicator of value.
Which brings us back to the headline promise. If SEO has an ROI problem and GEO has a grifter problem, what actually drives growth now?
The answer is not new. It has just been temporarily obscured.
What actually drives growth now
Strip away the tools, the acronyms, and the noise, and growth in search still comes from two places.
Capturing demand that already exists. Shaping demand before it does.
SEO earns its keep when it captures demand efficiently. It places you in front of people who are actively trying to buy something like what you sell. Its value can be approximated by replacement cost and customer lifetime value. If it brings in customers more cheaply than paid alternatives, it works. If it does not, it does not.
GEO earns its keep when it shapes demand. It influences what people and systems recommend when the choice is still being formed. Its value shows up indirectly, through increased brand search, higher conversion rates, improved close rates, and stronger preference.
Both disciplines have a role. Neither is magic.
The mistake is confusing the two.
SEO fails when it pretends to be brand building. GEO fails when it pretends to be optimisation. ROI thinking fails when it prioritises efficiency over contribution. Growth fails when marketing hides behind metrics that look rigorous but explain nothing.
The organisations that win in this next phase of search will be boring in the best possible way. They will do the maths on customer acquisition. They will invest in being known for something specific. They will fund influence where it actually matters and ignore dashboards that cannot be tied back to behaviour.
They will accept that some effects are measurable and some are inferential. They will stop demanding certainty where it does not exist and stop paying for theatre in its place.
This is not a new playbook. It is the old one, applied with more discipline.
What you need to take from this: the adult conversation search marketing has avoided
SEO does not need saving. It needs pricing honestly.
GEO does not need inventing. It needs defining properly.
SEO creates value when it gives you organic shelf space in places where people already spend money.
GEO creates value when it increases the likelihood that you are recommended when people and machines are deciding what to choose.
Everything else is distraction.
Traffic mattered, then it was abused, and now it matters again in fewer, more valuable places.
Citations sound impressive but mean nothing without influence. AI has not broken marketing. It has simply removed some of the hiding places.
The hard truth is this. Growth has never come from clever metrics. It comes from scale, relevance, and being remembered at the right moment.
Effectiveness first. Efficiency second. Placement for capture. Influence for choice.
If your SEO or GEO strategy cannot explain how it drives those outcomes, you are not optimising search.
You are wasting money.
In 2026, search is going to change, grow, and matter more than ever.
Don’t chase the wrong things.
Andrew Holland
Originally published on LinkedIn on 31 December 2025.