Latent Demand Is a Bill

Most growth problems are not marketing problems. They are targeting problems wearing a marketing costume. Spend goes up, results stay flat, and the instinct is to make better ads. The fix is almost always upstream: the wrong people, at the wrong moment, through the wrong channel.

We have written about the sequence that avoids this in build last and, for new markets, in when to expand internationally. This piece is about two specific ways the homework goes wrong, both of which are expensive and neither of which announces itself early.

ICP and persona are not the same thing

These get used interchangeably, and the confusion costs real money.

Your ideal customer profile describes the type of customer worth winning. For consumers that is a segment, defined by circumstance and spending behaviour rather than demographics alone. For business customers it is a type of organisation: a size band, a sector, a structure, a buying pattern. Written down properly it runs to two or three lines, and it should feel narrower than is comfortable.

A persona is a person inside that profile. In most deals there are several, and they want different things. In a referral or channel model the owner decides, the practitioner recommends, and the end customer pays and turns up. Each one needs a different message, and a value proposition written for the wrong one of the three will produce polite interest and no revenue.

This matters more in any model where you sell through someone else to reach the end customer, because you have acquired two buyers to convince rather than one. The upside is that you win the intermediary once and they keep sending people. The cost is that your messaging work doubles, and a partner who does not understand what they are recommending will send you the wrong customers with enthusiasm.

Existing demand, latent demand, and the bill attached to each

There are two kinds of demand, and the distinction decides the shape of your entire go-to-market.

Existing demand is people already looking for a solution to a problem they have named. They are searching, asking peers, comparing options and holding a budget. Your job is to be found, be credible and be chosen.

Latent demand is people who would benefit but do not know your category exists. Before you can sell your product you have to sell the idea of the product. That is called creating a market, and it is the most expensive game in marketing. You pay to educate everybody, including the people who will later buy a cheaper version from a competitor who spent nothing on education. It takes years rather than quarters, and the budget it needs is the kind a well-funded incumbent has and an early-stage company does not.

Plenty of worthwhile companies sit in latent demand. The point is to know which bill you have picked up, and to price it into the plan rather than discovering it in month nine when the ads have been running all year and the pipeline is full of curious people.

The funnel exists so that your message matches the moment

Marketers carve the buyer journey into top, middle and bottom of funnel. The jargon is unhelpful; the underlying discipline is not. Hitting a stranger with “book now” is the marketing equivalent of proposing on a first date.

Stage	What the buyer is asking	What works
Awareness	”Why is this happening to me, and is there a name for it?”	Articles, talks, explainers, anything that defines the problem rather than pitching the product
Consideration	”Would this type of solution help, and is it credible?”	Case studies, sample outputs, a free diagnostic conversation, evidence a sceptic can check
Decision	”Is it worth the price, and how soon can I start?”	Clear pricing, short time to value, references, a simple way to say yes
Retention	”Did it work, and what now?”	Follow-through, reporting back, a reason for the next step
Advocacy	”Would I tell someone?”	Reviews, introductions, outcome data the customer can share

Two things are worth noticing. The first is that a latent-demand business carries the entire awareness stage itself, which is the bill described above. The second is that most marketing budgets stop at Decision. In any referral-led business, Retention and Advocacy are where growth comes from, because a satisfied referrer sends the next ten customers at no acquisition cost.

The wrong customer looks like success at first

This is the part that catches experienced founders, because the early signals are encouraging. Clicks are cheap, forms fill up, the calendar of introductory calls is busy. Then very few of those people pay, or they pay and leave disappointed. Three recurring patterns:

People who want something adjacent that you cannot deliver. There is often enormous demand next door to your product for an outcome yours does not produce. Those leads convert badly, and when they do convert they arrive with the wrong expectation, which is worse.
Price comparators. If a cheaper version of something superficially similar exists, buyers who shop on price will find it. Marketing that attracts them burns money at the top of the funnel and loses at the bottom.
The interested with no budget. Happy to take the free assessment, the free call and the webinar. Rarely purchasers.

The remedy is not better creative. It is a narrower profile, and the discipline to measure cost per paying customer by source rather than cost per lead. Cost per lead is the metric that makes the wrong ICP look affordable.

Borrowing trust is cheaper than building awareness

Where demand is latent, the fastest route is usually to borrow somebody else’s trust rather than to manufacture your own. An intermediary’s customer already trusts the intermediary, already has the problem, and is already spending money trying to solve it. The intermediary does the educating in a ten-minute conversation, for free, with credibility you would take years to accumulate.

The trade-off is a familiar one. A paid channel starts immediately and stops the day you stop paying. A referral channel is slow to start and then compounds, and its true cost is your own time rather than media spend. Modelling both properly means counting the hours, including the founder’s, and comparing cost per paying customer rather than cost per enquiry. Most companies that run this comparison properly find the answer surprising, which is the argument for running it rather than assuming it.

In 2026, you are educating the machines as well

Here is what has changed since we first wrote about this. Creating a market used to mean teaching humans, one at a time, through channels you could buy. It now also means getting your category into the systems your buyers ask first.

Nearly all business buyers use AI somewhere in their buying process, and consumers increasingly start there too. When someone asks a model about a problem, it answers with the established category, because that is what the training data and the retrievable web contain. A product in a category that does not yet exist in those answers is invisible at the exact moment the buyer is forming their options, and no amount of advertising spend puts it there directly.

So the awareness bill now has a second line item. As we argued in cheaper to build, harder to be heard, being known has two audiences, and both reward the same thing: specifics that can be retrieved, extracted and checked. Published evidence, plain descriptions of what your product does and does not do, named outcomes, stated pricing. The work that makes you credible to a sceptical human is the same work that makes you legible to a machine, which is the one piece of good news in this article.

What to do about it

Before the next pound of acquisition spend:

Write the ICP down in two or three lines, and the personas inside it with what each one gains and fears.
Decide which kind of demand you are addressing. If it is latent, say so out loud, and either budget for the education or find an intermediary whose trust you can borrow.
Interview ten people about what they have already done and paid for, not about what they would hypothetically do. Past behaviour is evidence. Stated future intent is politeness.
Instrument attribution before you start, because you cannot retrospectively tag customers you did not track.
Set the number that would make you stop, and the date you will look at it.

None of this is new, and none of it is difficult. It is simply the homework, and as we argued in build last, the tools have removed the last excuse for skipping it. Being known beats being great, as we have said before. Being known to the right narrow group of people, at the moment they are already looking, beats both.