An ideal customer profile that is too broad produces three measurable effects: a message that resonates with nobody, an acquisition cost diluted across channels tested at random, and a cycle that stretches because reps qualify instead of converting. The right ICP is read from the customers who already succeed with you, and described by criteria observable from the outside.
Most of the companies we meet do not suffer from a shortage of market. They suffer from an excess of market, or more precisely from an inability to choose which one to address first. That indecision does not show up on a dashboard: it shows up in a rising acquisition cost, a lengthening cycle, and salespeople who spend their days qualifying.
“Anyone could be a customer”
That is the sentence that should alert you. When a founder tells us their product is for “any company that wants to save time”, they are describing a dream, not a market. An ICP that is too broad produces three immediate consequences, and they reinforce each other.
- A message that resonates with nobody. To speak to everyone, you end up saying nothing precise. The prospect reads a generic promise and does not recognise themselves in it.
- An acquisition cost that explodes. With no clear target, every channel gets tested just in case. Budget scatters, and no channel ever receives enough to prove anything.
- A sales cycle that stretches. Reps spend their time qualifying instead of converting, and deals that should never have entered the pipeline sit there for months.
The three measurable symptoms
Before discussing definitions, measure. Three numbers reveal an overly broad ICP better than any workshop.
The spread of your won deals. If your last twenty customers sit in fifteen different sectors, you do not have an ICP, you have a history. Companies that win almost always show an abnormal concentration somewhere, and they have usually not noticed it.
The conversion gap between segments. Slice your pipeline by size, by sector, by source. A threefold difference in yield almost always appears, and it points at the segment you should be serving first.
The churn rate by profile. The customers who leave fastest are often the ones who should never have signed. If your departures concentrate on one profile, reality is already dictating your ICP to you.
Narrowing is not giving up market
The fear is always the same: “if I focus on one segment, I lose the others.” The opposite happens. A sharp position on a priority segment creates a reputation, comparable references, word of mouth inside a single community. Those proofs are what later open the adjacent segments, and they do not exist when you have served everyone by halves.
There is also a mechanical effect on cost: a tight segment makes signals observable, communities identifiable and channels predictable. That is the precondition for prospecting triggered by events rather than by a list, as we describe in signal-based selling.
The three axes of a workable ICP
A usable ICP is defined on three axes, and all three have to hold. A segment that satisfies only two produces deals that drag.
The sharpest pain. Who has the problem you solve best, badly enough to pay now? Not “who agrees it is a topic”, but who has already built an internal workaround, hired someone for it, or missed something because of it.
The ability to buy. An existing budget, an identifiable decision-maker, a cycle that is realistic given your cash position. An enthusiastic segment that takes eighteen months to decide is not your first segment, even if it is your dream market.
The ease of reaching them. Identifiable channels, signals observable from outside, an existing community, a directory or database available. A segment you cannot reach at a reasonable cost remains an intention.
Writing an ICP that actually serves
An ICP is described by criteria observable without talking to the prospect. That test eliminates half the definitions we see. “Companies that care about data quality” cannot be targeted: you cannot know from the outside who ticks that box. “Software vendors of 50 to 200 people who have just hired a revenue operations lead” can be counted, listed and worked on Monday morning.
Always add a triggering element of context, because that is what sets the timing. A funding round, a leadership change, a tooling migration, a new regulatory obligation. Without a trigger you have a target but no calendar.
Finally, write down what the ICP excludes, explicitly. A definition that excludes nobody guides no decision, and exclusion is where the quality of the exercise is measured.
The Arkimoz method
We always start from the customers who already succeed with you. Not the ones you dream of: the ones who renew, refer, and talk about you. That core reveals a pattern nobody had formalised, and that pattern is your real ICP, not your theoretical one.
We then cross that core with your lost deals, which often teach more than the won ones: they say where your promise does not hold, and why. We formalise the definition with its observable criteria and its trigger, write the message that speaks to that segment, and align acquisition behind it. Everything downstream, the promise and the sales motion, follows from that decision, which we place in context in building a go-to-market plan.
Where to start this week
Take your last twenty won deals and your last twenty lost ones. Put them in a table with five columns: size, sector, business model, source, and what was happening at their company at the time of the first conversation. A concentration almost always appears on two of those columns.
That single table settles more than three positioning workshops, and it gives you the material to write the promise that follows. The first euro of profitable growth never comes from broadening. It comes from precision.
Frequently asked questions
- What is an ICP in B2B?
- The ICP, or ideal customer profile, describes the type of company where your offer creates the most value and sells best. It describes the company, not the person: size, sector, business model, maturity, triggering event. The profile of the individual you talk to is the persona, which is a separate layer.
- How do you define an ICP with very few customers?
- With ten customers, retrospective analysis already works, as long as you add the deals you lost. Look for what separates the three customers who get a result from the three who do not. The difference is rarely the sector, and often a piece of internal context such as a dedicated team or an available data set.
- Is an ICP that is too narrow a risk?
- It is a theoretical risk and a rare one in practice. The real case to watch is a segment too small to carry your targets: count the companies matching your criteria and compare that number with your revenue goal and your conversion rate. If the arithmetic does not work, widen on one axis, not on three.
- How often should you revisit your ICP?
- Once a year, and whenever a signal appears: a lengthening cycle with no external cause, a conversion rate dropping on one segment, or customer churn concentrating on one profile. An ICP is a working hypothesis that reality corrects, not a carved definition.