The ICP test: ask three people on your team who the ideal customer is

The three-colleague experiment

It's almost always the first thing I do when I join a company, because it's fast, free and brutally revealing. The three people answer separately, so nobody lines up behind anyone else. And the answers almost never match: sometimes they don't even vaguely match.

The founder says "SMEs that want to grow". The salesperson, in practice, only closes companies above a certain size with a certain role already in place. Marketing targets startups. Three answers, three different companies, all served badly at the same time. And nobody had noticed, because everyone was looking only at their own piece.

The value of the test is in its discomfort. It lines up three partial truths and brings out the inconsistency. From that moment you can't pretend any more that "we all know who the customer is": it's written in front of you that it isn't true.

Why three different answers are a system problem

If there are three internal definitions of the ideal customer, you have three engines pulling in three directions. Sales qualifies with one criterion, marketing generates leads with another, the founder describes a third on calls. The net result is dispersion: effort that cancels itself out.

The subtle point, the one that makes a muddled ICP so dangerous, is that it doesn't show up as confusion. It shows up as symptoms that look independent: the message that changes on every call; deals that stall because you're talking to the wrong decision maker; marketing bringing leads that sales throws away; a value proposition that sounds different on every channel; "low conversion" that nobody can explain. They look like five problems to solve separately. It's one, upstream.

That's why treating the symptoms doesn't work. Rewrite the message? It will change again on the next call, because there's no agreement on who it's for. Train sales on qualification? They'll use different criteria from marketing. Until the three heads converge on a single definition, every downstream fix is temporary.

ICP vs buyer persona: not the same thing

A frequent confusion, worth clearing up because it leads to analysing at the wrong level.

The ICP describes the ideal company to serve: industry, size, business model, maturity, the situation that makes it a good customer. It's the filter that decides whether a company is worth contacting.

The buyer persona describes the person inside that company you talk to: role, priorities, objections, what makes them decide. It's the filter that decides how you talk to them once you've decided the company is right.

Order: first the ICP (who we sell to, at company level), then the buyer persona (who we talk to inside that company). Skipping the ICP and starting from people is the most common way to build crowded, imprecise targeting.

How to build an ICP from real customers (not projections)

The move isn't "defining the ICP" in the abstract on a strategy slide. It's starting from a fact people often refuse to look at: who you already actually close, and who stays.

Take your best customers, not the biggest, the best: the ones who close fast, stay, pay without friction, refer others. Look at what they have in common. Not what you'd like them to have in common: what they actually share. Industry, size, the moment they were in when they bought from you, the role of whoever decided, the specific problem that pushed them. The pattern that emerges is your real ICP, written by the data and not by wishes.

In the anonymised example above (the company with three different answers and "low conversion") the fix was exactly this: we took the segment sales was really closing, the one with real data behind it, and aligned marketing and message on it. Conversion went up not because we "worked on conversion", but because everyone was finally selling to the same person.

The ICP isn't an aspiration. It's a fact. And the fact is often smaller and duller than the founder would like, but it's the one that converts.

What changes downstream when the ICP is clear

When the three heads converge, the system stops wasting energy. Targeting narrows and improves: smaller, righter lists beat huge, wide ones, every time. The message settles, because everyone knows who they're talking to. Qualification becomes consistent between marketing and sales. And "conversion" rises as an effect, not as a goal chased directly.

This is where strategy stops changing on every call: when there's a single sentence (this customer, this problem, this reason-us) and everyone can repeat it without translating it.

A team that gives three different answers about the ideal customer doesn't have a conversion problem. It has three companies trying to live inside one. The job is picking one: the one that already works.