Why Great Online Marketing Is Not a Defensible Business Model
What you’ll get from this piece, and the problems it helps you solve.
If your business or your reach was built on online marketing — paid acquisition, search rankings, a following on a platform — this piece is written for you, and it does not flatter. By the end of it you’ll be able to look at your own business and answer a question most founders never ask cleanly: is what protects me actually mine, or am I renting it? You’ll have a precise way to tell the difference between a real competitive advantage and a borrowed one that can be repriced, copied, or switched off by someone you don’t control.
Concretely, you’ll come away able to do four things. Diagnose where your customer relationship actually lives — on your own ground, or on a platform’s. See how a single dominant channel quietly rewrites your whole business model without anyone deciding it should. Understand why the current scramble to “do online marketing for AI” is the same mistake in newer clothing, so you can stop reflexively renewing the lease. And recognise where durable defensibility genuinely comes from — so you can start building the kind that no algorithm update can take from you.
It will take fifteen minutes and it may sting. That’s intended. The discomfort is the point at which the thinking starts.
This is the piece behind the manifesto, I just published simultanously — the argument under the shout. It is uncomfortable on purpose, because skipping the discomfort is itself a kind of bad strategy. And bad strategy, it turns out, has a precise definition.
Richard Rumelt opens Good Strategy / Bad Strategy with a demand that sounds obvious and is almost never met: before anything else, diagnose. The fundamental task is comprehending the situation and identifying the biggest barrier to forward progress — not naming a goal, not reciting a vision, not reaching for a slogan. Jim Collins frames the same discipline as confronting the brutal facts while keeping faith that you’ll find a way through. Most companies do neither: they skip the brutal facts and keep the faith anyway. That’s not strategy. That’s hope in a strategy costume.
So let’s diagnose. Not the whole economy — a specific population. This is about online-native businesses and creators: the direct-to-consumer (DTC) brands, the pure-play shops, the affiliate and comparison sites, the content publishers, and the influencers who built their entire success on online marketing. People and companies whose reach, revenue, and relationship with their audience were constructed, from day one, on top of someone else’s platform.
The brutal fact, stated plainly: you don’t own your customer relationship
They pawned it, one quarter at a time, to platforms they don’t control — and many still call the result a competitive advantage. It isn’t one. A relationship you can’t reach without paying a third party for access is not an asset on your books; it’s a line of credit the lender can call at any time.
The lender is calling it now, because AI is pulling the foundation out in real time. In the year to November 2025, Google search traffic to publishers fell by roughly a third globally, and zero-click searches — where the user gets an answer and never visits a site — now make up the majority of queries. Individual sites have been hit far harder: HubSpot has estimated it lost 70 to 80 percent of its organic traffic. On the paid side, the bill keeps climbing: ecommerce customer acquisition costs are up roughly 40 percent since 2023, with one analysis putting the rise at 222 percent over eight years.
These are not marketing fluctuations. They are the invoice for a business model decision — one that, in most cases, nobody ever consciously made. That’s the diagnosis. Everything below is the evidence, and the explanation.
Reading it through the business model lens: one channel quietly takes over all four elements
This is where I have to insist on my own discipline. I don’t analyze marketing. I analyze business models. And a business model in my framework is not a list of parts. It is a system of four interdependent elements: the value proposition (what value you promise to which real human being), the value-creation architecture (how you produce and deliver it, with whom, and how you reach the customer), the revenue model (how the money actually flows), and the Unternehmensgeist — a term I built with the French esprit in mind, meaning both the mind and the soul of a thing: the animating spirit of the enterprise, its people, its values, and its reason for being.eople and its reason for existing.
Customer access — how you reach the people you serve — is one component inside the value-creation architecture. One component. The pathology of an online-marketing-built business is that this single component quietly takes over the whole system. Watch what happens to the other three when it does.
The value proposition stops being shaped by the customer and starts being shaped by the channel. You no longer build what a real person deeply needs; you build what ranks, what the feed rewards, what the ad auction favors. The promise is made to the algorithm, not the human.
The revenue model gets calibrated to an acquisition cost that a third party sets and can raise at will. Your unit economics are no longer yours; they are a function of someone else’s auction.
And the Unternehmensgeist — the esprit, the animating soul — slowly dies. This is the cruelest part, because it is the hardest to win back. A team that worships the channel has handed its spirit to a false god: an idol that gives nothing back, demands constant sacrifice, and means nothing. People reorganize around dashboards instead of around customers. They stop asking why the company exists and start asking how to rank. The very thing the French word points to — the soul of the enterprise — is precisely what the idol of online marketing consumes first. And a company that has forgotten its reason for being has nothing left to be unique about.
How it happens: nobody decides to do it — it emerges one quarter at a time
Here is the fairest thing that can be said about the companies in this trap, and also the most damning: nobody decided to do this. There was no board meeting where someone proposed abandoning the value proposition. It emerged.
Henry Mintzberg drew the distinction decades ago between deliberate strategy — what you intend and write down — and emergent strategy — the pattern that actually forms out of a stream of individual decisions. What a company truly does is rarely what its strategy document says; it’s the pattern that emerges from a thousand small, locally sensible choices. And that is exactly how the inversion happens. This product, because it ranks. This headline, because it converts. This feature, because the search volume is there. Each choice is defensible on its own. The pattern they form is surrender — and no one ever chose the pattern.
What pulls the emergence in the channel’s direction rather than the customer’s? The measurement system. The platform produces the loudest, cleanest, most immediate signals any business has ever had: clicks, rankings, return on ad spend. An emergent strategy follows the signal it can measure, and the channel shouts its signal in real time while the real customer — slow, qualitative, hard to quantify — barely whispers. So the strategy emerges toward the thing that’s easy to count.
This is where the modern management religion does its quiet damage. Agile, lean, sprints, OKRs — these are superb instruments for one kind of problem, and a disaster mistaken for a worldview. They are built for what the Cynefin framework calls the complicated: a domain of known cause and effect, where fast iteration against a clear metric is exactly right. They are blind, by construction, to the prior question — whether the metric still points at anything that matters. The sprint asks “are we building it right?” It never asks “are we building the right thing?” And because the cadence rewards what moves inside a short cycle, the long, unmeasurable, generative things — a brand, a relationship, a reason to exist — get crowded out. Not by malice. By rhythm. A fool with a tool is still a fool; an entire company with excellent tools and no memory of why is simply sprinting more efficiently in the wrong direction.
Why it’s a trap, not bad luck: optimizing the click means optimizing the past
Clayton Christensen’s Innovator’s Dilemma showed that well-run companies often fail precisely because they do what good management says: they listen to their best current customers and relentlessly optimize their existing processes. The online-marketing-built business is the purest expression of that logic — and one turn worse. It doesn’t only listen to today’s customers. It listens to today’s clicks. And the one who clicks need not be a customer at all. A search query is a record of what people already know to ask for. It is a map of the past. Build only what the volume validates and you will produce a faster horse forever, with magnificent efficiency, right up until the car arrives.
That is the efficiency trap in one line: efficiency measures how well you do something. It is silent on whether the something is still worth doing. A metric measures an assumption about the world — the click assumes there are clicks, that they reach a customer, that the platform keeps sending them. When AI answers the query on the page, the assumption dies, but the dashboard keeps counting, flawlessly, an event that no longer occurs. There is nothing more efficient than the wrong answer, perfectly executed.
Two illustrations: the influencer and the DTC brand rest on a single rented block
Take the influencer or creator first, because it is the cleanest case there is. A creator with millions of followers feels like the owner of a thriving business. Read it through the framework and the illusion dissolves. The value-creation architecture is a single borrowed component: an audience that lives on a platform and is governed by an algorithm. The relationship was never owned; it was rented, and the rent is paid in constant content. When the platform changes its ranking logic — or an AI feed reshapes what gets seen — the audience can evaporate overnight, and there is nothing underneath to fall back on. One strong component on someone else’s land is not a business model. It is a tenancy.
Now the direct-to-consumer brand. The original DTC promise was disintermediation: skip the retailer, own the customer. But “own the customer” turned out to mean “reach the customer through Meta and Google,” and that toll rises every year. When a brand spends up to half its marketing budget on increasingly expensive paid ads, the platform — not the brand — captures most of the margin that disintermediation was supposed to create. The revenue model was quietly rewritten in the landlord’s favor, and most brands only noticed when the rent became unpayable.
The false exit: ‘online marketing for AI’ is the same mistake in newer clothing
Watch what the industry is doing now that the foundation is visibly crumbling. It is shouting for new clothes. Generative engine optimization. LLM visibility. “How do I rank in ChatGPT?” The exact reflex that created the dependency, aimed at the next platform — one that won’t even tell you why it cited you or didn’t.
By Rumelt’s own definition, this is bad strategy in textbook form. Bad strategy is fluff — vague language pretending to be expertise — and the failure to face the actual challenge, confusing a goal with a strategy. “Be visible in AI” is a goal dressed as a plan. It diagnoses nothing. It changes nothing about the underlying structure. It simply swaps one landlord for a newer, more opaque one and calls the move innovation. You are not escaping the trap. You are renewing the lease.
The concept behind it all: the Uniqueness of your Business Model (UBM)
Here is the idea I want every reader to leave with, because it reframes the entire problem.
We all know the USP — the unique selling proposition, the one distinctive thing you say about a product. The USP is a single building block. What actually defends a business is something larger, which I call the UBM: the Uniqueness of your Business Model. And the crucial word is of the business model — not of one element. A UBM is uniqueness that arises from the interplay of several components: a value proposition, an architecture, a revenue model, and a spirit that reinforce one another into a configuration a competitor cannot easily disentangle and a platform cannot simply switch off.
This is why “brilliant online marketing” can never be a defensible position. At its best, it is one strong building block. And a single building block — especially one built on rented ground — is by definition not a UBM. It can be copied (a competitor buys the same keywords), removed (an algorithm update), or repriced (the auction climbs). Defensibility never lives in one component. It lives in the interdependence between several that you actually own. The influencer with five million followers has no UBM — one borrowed block on someone else’s land. The AI disruption didn’t destroy these business models. It revealed that, in terms of defensibility, there was never much of a business model there to begin with.
A moat you don’t own is not a moat. It’s a tollbooth.
The real question: serve search volume, or build on genuine customer insight
So we arrive, finally, at the question the manifesto ends on — and it is a genuine fork, not a rhetorical flourish:
Do you serve the customer who arrives through a search box, or do you have the nerve to believe you offer something better than search volume — something built on a real understanding of a real human being?
One answer is analytical: it describes what already exists, the demand already articulated, the volume already there. It can only ever produce variations on the present. The other is generative: it asks about the unspoken need, the job the customer can’t yet name, and from that it can create something nobody is searching for yet — because it doesn’t exist yet. Search volume would never have produced the thing nobody knew to want. The capacity to see a future clearly enough to build it is precisely the capacity that worshipping the channel switches off.
This is the Stockdale discipline made concrete. The brutal fact is that a business resting on rented access is undefended. The faith is that there is a way through — but it runs through re-owning the customer, not through finding a cleverer landlord.
Lessons learned: check whether your uniqueness is owned or rented
1. Defensibility lives in the UBM, never in a single block. Ask of your own business: does my uniqueness come from the interplay of several components I own — or from one strong element, possibly a rented one? If a single algorithm change, auction, or policy update could remove it, it was never a moat. It was a tollbooth you mistook for one.
2. Know which of your channels you own and which you rent. An owned customer relationship sits inside your own architecture; a rented one is a cost on someone else’s terms. Map yours before the landlord raises the rent or changes the locks — because they can, and eventually they will.
3. Watch for a single component swallowing the system. When your value proposition, your pricing, and your team have all reorganized themselves around one channel, that channel is no longer a part of your business model — it has become its master. Concentration disguised as competence is the most dangerous vulnerability there is.
4. Efficiency is not the same as being right. A channel can deliver an excellent return and still be the wrong foundation. Measuring how well you do something will never tell you whether it remains worth doing. Build the second question into your rhythm deliberately, because no dashboard will ask it for you.
5. When the ground shifts, diagnose — don’t reach for the next buzzword. The collapse of a channel is not solved by a faster version of the thing that caused it. “Online marketing, but for AI” is a goal pretending to be a strategy. The honest move is Rumelt’s: name the real challenge — undefended customer access — and rebuild the interdependence that makes a business model genuinely its own.
This analysis uses my four-element business model framework— value proposition, value-creation architecture, revenue model, and Unternehmensgeist — and the concept of the UBM, the Uniqueness of your Business Model. The recurring lesson holds here as everywhere: durable advantage comes not from any single strong element, but from the interdependence between them. That is the whole reason I wrote the manifesto — read it here.
