The company is buying the skill of harvesting a field that has not been sown.
I have spent the past few days reading fractional CMO roles in B2B, and in almost every second one I kept tripping over the same combination. The products themselves differ: an AI platform, complex fintech, embedded payments for non-financial services, cybersecurity. What they share is their stage. These are new categories that did not exist in this form even five years ago, and the market does not yet quite understand why it needs them. And yet the brief for the marketer is always the same: set up the funnel, the landing pages, the performance campaigns. As someone who studied marketing back when it still lived in textbooks, I trip over this every time, and I decided to work out why.
The short answer: the company is buying the skill of harvesting a field that has not been sown. The longer answer is below, and it is worth ten minutes if you are bringing to market a product the market has not yet asked for.
01 / PERFORMANCE DOES EXACTLY ONE THING
Let us settle the terms first, because it is the blurring of them that creates the confusion.
Performance marketing does one thing, and does it well: it finds people whose need is already formed and walks them to a purchase. If someone types embedded payments provider into a search bar, they already know they want an embedded payments provider. Performance intercepts them: it shows the ad, sends them to a landing page, captures the contact. It is quick and it counts down to the last click, so the dashboard shows the number of leads, their cost, and the conversion rate. This is the skill companies buy most readily. The finance director sleeps soundly.
The problem is not performance. The problem is the type of demand it is applied to.
02 / TWO BUYERS WHO GET CONFUSED FOR ONE
In the marketing of complex products, it is worth distinguishing two quite different types of buyer.
The first is in-market. They have recognised the problem, started looking for a solution, are comparing options and are ready to talk to a salesperson. A landing page does genuinely catch these people, and for them performance is the right instrument.
The second is out-of-market. They are not looking for anything, because they do not believe they have a problem that can be solved in this way. Not because they are inattentive, but because such a solution did not exist in their picture of the world. For a genuinely new product, these people are the majority of the market, and no landing page will ever reach them: you cannot catch someone in a funnel they never enter.
This is where the fault line runs. The company sells a product whose buyers sit largely out-of-market, and hires a specialist who can only work with those already in it. And there are few of those, because the category is new. The marketer honestly works the available demand, hits its ceiling within a couple of quarters, and everyone is left wondering why the marketing isn't working.
03 / WHAT DEMAND GENERATION ORIGINALLY MEANT
The term that describes the work with the second type of buyer has been around a long time. Demand generation. The trouble is that over the past fifteen years it has been hollowed out into a synonym for performance. Today, in job postings, demand gen usually means the same leads and forms.
Originally it meant precisely the opposite: to create demand where there is none yet. To show the market that a familiar problem can now be solved in a new way that did not exist before. The hard part here is not bringing in traffic. The hard part is explaining an unfamiliar, complex thing clearly enough that a person recognises their own problem in it.
And the mechanics of this work have not changed since before the internet. Long before digital channels, demand for a complex product was created through industry reports, reference clients, seminars, columns in trade publications, public speaking. Today case studies, expert articles, plain explanation of how the product works, and guidance on how a client should actually use it have been added to that list. The channel has changed, not the mechanics.
Demand for a complex product is still created not by a form on a website, but by a case with someone else's problem solved and a reputation trusted before the first contact.
The difference between the two kinds of work is not cosmetic. Performance answers the question how do we sell to those who already want it. Demand generation answers the question how do we make those who do not yet know it is possible want it at all. These are different tasks, different skills and, as a rule, different people. Alice de Courcy, former CMO of Cognism, puts it more bluntly: you have to move beyond a lead-driven mindset and focus on creating demand among the near-everyone who is not yet in-market. The thin layer of ready buyers takes little time to gather; the real work begins after that.
04 / WHOM DEMAND IS ACTUALLY CREATED ON
A fair question: if most of the market is out-of-market, where does one begin? Not with everyone at once. That is both impossible and ruinously expensive.
Here the old diffusion of innovations model, described by Everett Rogers in 1962, is useful. A new product spreads not evenly but in waves: the first to take it up are not everyone, but a narrow group of early adopters, those willing to try the unfamiliar ahead of the rest because it matters to them to be first in their niche. Demand is created on them first. They become the very reference clients whose example is then used to explain the product to the next, more cautious layer of the market.
An important caveat, described in detail by Geoffrey Moore: between these early adopters and the so-called early majority lies a gap. What convinced the first does not automatically convince the majority, and the reason is telling. Early adopters decide on the merits of the product itself, while the early majority leans far more on the reputation of whoever makes it. That crossing deserves its own discussion; precisely whom to move across the gap, and how, I will take apart another time. Here it is enough to fix one thing: creating demand begins not with an attempt to reach everyone, but with a correctly chosen first layer.
05 / NOW ADD AI TO THIS
And onto this already unbalanced picture comes a change in the very way the buyer searches.
More and more often the buyer starts not with a search engine but with a question to a language model: what do I need, whom would you recommend. And they receive a short list of vendors before any first contact with a single one of them. Something quiet but serious happens: the funnel of discovery turns, in effect, upside down. You used to be found early, while the person was still feeling out a solution and weighing options. Now the model surfaces you late, when the buyer already knows the name and is asking specifically about you. Those still to be discovered it usually does not add to the list at all.
And the model does not take that name from an advertising budget. It assembles it from what is confirmed across the web: case studies, mentions, an expert voice, a clear sense of what you actually do. If the model does not consider your brand an authoritative answer to the query, you lose the deal before sales ever hears of it. In essence this is the same pre-internet arsenal of demand creation, reputation, clarity, third-party confirmation, only now a machine reads it too.
The irony is that this strikes at exactly the skill companies buy first. Performance gathers those who have already matured. But the matured are becoming fewer, because part of the choice now happens on the AI's side, at the very entrance, before any click on an ad. And the contribution of reputation cannot be measured by the usual last-click attribution: the click you could assign simply is not there in this chain.
06 / WHAT TO DO WITH ALL THIS
I am not proposing to bury performance. It is necessary, and for a mature category with ready demand it remains the best instrument there is. I am proposing to call things by their proper names.
If you are bringing to market a product the market has not yet asked for, performance is a necessary but insufficient condition. It will gather the thin layer of demand that already exists and run up against its limit. Beyond that begins different work: explaining the category, creating demand among those still out-of-market, and building a name that not only a person but a language model will surface. That work is, as a rule, nowhere in the brief.
Performance harvests. But someone has to sow the field first, and make sure there is something to harvest a year from now. That is the line worth adding to the role first.
If this logic resonates and you are weighing what kind of marketing your product needs at its current stage, that is precisely the question I begin with when I work with a company.