What Is an ICP? How to Define Your Ideal Customer Profile and Choose a Beachhead Segment
What an ideal customer profile is, how it differs from a buyer persona and a target market, how to define one before you have revenue, and how to pick the one segment to win first.
By Suren, Founder & Growth Strategist at Gufy · Published
ICP stands for ideal customer profile (some people say ideal customer persona): a description of the customer who gets the most value from your product and is most worth your effort to win. For a startup, the useful version is narrow. It names one segment you can reach, who feel the problem badly and can pay, and it says who you will not chase yet.
This guide explains the term, shows how it differs from a buyer persona and a target market, and walks through defining one when you have few or no customers, including how to choose the one segment to start with.
What an ICP is, and what it is not
| Term | What it describes | Example |
|---|---|---|
| Target market | Everyone who could plausibly buy | Small professional-services firms |
| Ideal customer profile (ICP) | The type of customer you should win first, and why | Law firms with 5 to 20 fee earners that still review documents by hand |
| Buyer persona | A person involved in the purchase | The principal who approves software spend and worries about risk |
| Beachhead segment | The single ICP you concentrate on until you lead it | The same law firms, in one practice area and one country |
In business software the ICP is a kind of company. In a consumer app it is a kind of person in a particular situation. Either way it describes a customer, not a demographic average.
Why "everyone" fails
On founder calls, the most common answer to "who is it for?" is some version of "anyone who…". It is understandable. The product could help many people, and choosing feels like turning business away.
Three things go wrong when you do not choose:
- The message goes vague. Copy that has to fit every customer describes none of them. A narrow profile is what makes a specific positioning statement possible.
- The budget spreads thin. Reaching three audiences with a third of the money each usually means not enough evidence from any of them.
- You cannot learn. If ten sign-ups come from ten different kinds of customer, you have ten anecdotes and no pattern.
One founder we spoke with was launching a marketplace and wanted every kind of buyer, seller and professional on it from day one. Even "buyers" was several customers: someone buying a first home, an investor and a developer looking for a site want different things and respond to different messages. Put them in one bucket and the product, the copy and the first-time experience all blur.
Deep experience makes this harder, not easier. A founder with decades in an industry sees its problems so vividly that every one of them looks worth solving at once. The experience that is your advantage is also what stops you choosing. The fix is to get it out of your head and on to paper, one customer type at a time, then pick.
There is a third reason to choose. Building software keeps getting easier, so a competitor can copy your features. What they cannot copy quickly is how well you understand one particular customer.
What goes into an ICP
A usable profile fits on one page.
| Part | What to write | Example (legal software) |
|---|---|---|
| Who they are | Industry, size, location or, for consumers, life situation | Law firms, 5 to 20 fee earners |
| The trigger | What has happened that makes them look now | Lost a tender on turnaround time; a senior associate left |
| The problem | The job they are trying to get done, in their words | "Reviewing contracts eats our juniors' week" |
| What they do today | The current workaround and what it costs them | Manual review, with a partner checking everything |
| User and buyer | Who uses it, who pays, who can block it | Associates use it; the principal buys; IT or risk can veto |
| Ability to pay | Evidence they spend on this kind of thing | Already pay for practice-management and research tools |
| How to reach them | Where they can be found | Industry associations, LinkedIn, referrals between firms |
| Not for | Who you will turn down for now | Sole practitioners; firms with an in-house innovation team |
User and buyer are often different people
A product can delight the person who uses it and still not sell, because someone else signs it off. A founder building an AI tool for lawyers had tested the idea with lawyers he knew. The people who would approve the purchase were firm principals, whose concerns were billable hours and risk, not the technology.
Write down both, and make a guess for each: what would make the buyer give you half an hour, and how does the user's working day change if the firm adopts it? Your product has to win the user. Your message and pricing have to win the buyer.
Write the "not for" list
An anti-ICP is the list of customers you will not pursue yet: too small to pay, too large to sell to without a sales team, too slow to decide, or needing features you do not have. It saves more time than any other line on the page, because it gives you permission to say no to a lead.
How to define an ICP before you have revenue
Most guides tell you to analyse your best customers. That does not help if you have ten free users or none. Before revenue, the ICP is a hypothesis, and the work is choosing the best one and testing it.
1. List the candidate segments
Write down every group that could use the product. Be specific: "physiotherapy clinics with two to ten staff", not "healthcare".
2. Score each one
Give each segment a score from 1 to 5 on five questions.
| Question | What a 5 looks like |
|---|---|
| How painful is the problem? | They already spend time or money trying to fix it |
| Can they pay? | They buy similar tools and have a budget for it |
| Can you reach them? | You can name where they gather and get in front of 50 of them |
| Do you know them? | You have worked in the industry or have a network in it |
| Do they talk to each other? | Referrals and word of mouth are normal in the group |
The fourth question is the one founders underrate. A founder with a product for "all small businesses" had ten unpaid users. The three who ran their whole business on it all came from the one industry she had worked in herself, because she had been able to show them exactly how to use it. A salon, an accountant and a plumber do not think about admin the same way, and a pitch built on one trade's experience falls flat with another. Her own industry was the obvious place to start: she knew its problems, spoke its language and could reach twenty or thirty owners directly.
Our standing advice is to start with an audience whose problems you have lived. You already know what the solution has to do.
3. Pick one, and write the profile
Take the highest-scoring segment and fill in the table above. Leave blanks where you are guessing and mark them as guesses.
4. Test it with five people who do not know you
People in your network will be kind about your idea. A founder whose evidence was three friendly lawyers needed strangers before building further. Five conversations with people in the segment who have no reason to be polite will tell you:
- whether the problem is real and how they describe it;
- what they do about it now;
- who would have to approve a purchase;
- whether they would pay, and roughly what the outcome is worth to them. Unit Economics for Startups shows how to turn that into the most you can afford to spend on winning a customer.
Record the conversations if they agree. You will hear the answers you were hoping for, and someone else listening later will hear different things.
Pay attention to how many of the five actually want what you are offering. If one says "yes, take that task off my hands" and four say "that task is the part of my job I value", your real market is the one in five, and the profile needs to describe that person and not the other four.
It also helps to get one customer using the product closely, even free, as a pilot. What they ask for gives you the question to put to the next prospect: "another firm like yours needed this. Is it a problem for you too?"
A survey can show that a problem exists. It rarely shows that people will pay to solve it, so ask for a commitment: a pre-order, a paid pilot or a discounted early sign-up before launch. Until a stranger pays you, everything in the profile is a hypothesis. The first payment from someone with no relationship to you is the earliest real sign that you have the right customer. The SaaS idea validation playbook covers the interview questions and what counts as evidence.
5. Win the first twenty to thirty paying customers by hand
Before running ads, get the first paying customers from the segment through people you can reach directly: your network, introductions, direct messages, demos you run yourself. Twenty to thirty paying customers of the same type is a pattern. It also gives you the retention and referral evidence that an ICP is supposed to rest on. Early-Stage Marketing for Startups describes how to find them.
6. Revise it from what you see
Once people are paying, compare the profile with reality. Which customers reached value fastest, stayed and referred others? Tighten the profile around them. If your product has a trial, look at who reaches the aha moment and who does not: the customers who activate easily are usually your ICP.
Choosing a beachhead segment
The term is military. In the Normandy landings of 1944 the Allies did not try to take the whole coast of Europe at once. They took a few beaches, held them, and expanded from there. Land, then expand.
For a startup, the beachhead is the one ICP you serve first, which means not catering to anyone else at the start. Trying to serve anyone and everyone is the most common mistake we see early founders make, and those who learn it usually learn it the expensive way.
The idea was brought into business by Geoffrey Moore's Crossing the Chasm and developed for startups by Bill Aulet at MIT in Disciplined Entrepreneurship. The principle is to concentrate on one segment small enough to lead, then use that position to expand.
Aulet gives three conditions for a segment to count as one market:
- The customers buy similar products.
- They have a similar sales cycle and expect value in similar ways.
- There is word of mouth between them.
The third is what makes a beachhead work. In a segment where customers know each other, every sale makes the next one easier. Spread the same number of sales across unrelated groups and you get no such effect.
A good beachhead is smaller than feels comfortable. If you can list the first hundred target customers by name, it is about the right size.
For a marketplace, choose a beachhead on each side separately, then check that the problem you are solving lines up across them. A narrow group of buyers is only useful if the sellers you recruit have what those buyers want.
A beachhead also helps if you plan to raise money. Early investors look for traction, and traction comes sooner from one segment you understand deeply than from five you have touched lightly.
A worked example
Imagine staff-rostering software for hospitality.
- Target market: hospitality businesses.
- Candidate segments: hotel groups; franchise outlets; independent cafés and restaurants with 10 to 40 staff; event caterers.
- Scoring: hotel groups have systems already and long sales cycles. Franchise outlets use whatever head office chooses. Caterers roster around events, which is a different problem. Independent venues build the roster by hand every week, lose hours to swaps and no-shows, decide quickly, and their owners know each other.
- Beachhead: independent cafés and restaurants with 10 to 40 staff, in one city.
- Not for, yet: hotel groups, franchises and caterers.
- Next segments once this one is won: neighbouring cities, then bars and small hotels with the same weekly problem.
A useful check on the size: a founder who can say "my focus is businesses of this type, below this size, and I am not selling to anyone else for now" has a beachhead. One who says "it could also work for…" does not have one yet.
Common mistakes
- Analysing from the outside. A profile built from what you assume about an industry you have not worked in needs conversations before it needs marketing.
- Describing a demographic, not a customer. "Women aged 25 to 45" tells you nothing about who has the problem.
- Choosing the largest segment. Size matters later. First you need a segment you can win.
- Building it from friends' opinions. Use strangers and, as soon as you can, payments.
- Confusing the user with the buyer.
- No "not for" list.
- Writing it once. The profile should change as paying customers teach you who they are.
Common questions
What does ICP stand for?
ICP stands for ideal customer profile. It is a description of the type of customer that gets the most value from your product and is the most worthwhile for you to win: they have the problem badly, can pay, can be reached, and tend to stay.
What is the difference between an ICP and a buyer persona?
An ICP describes the type of company or customer you should sell to. A buyer persona describes a person involved in the purchase: their role, goals and objections. In business software you define the ICP first, then one persona for the person who uses the product and one for the person who approves the spend.
How do you define an ICP with no customers?
Treat it as a hypothesis. List the groups that could use the product, score each on how painful the problem is, whether they can pay, how easily you can reach them and how well you know them, pick one, and test it by interviewing five people in that group who do not know you.
What is a beachhead market?
A beachhead market is the single, narrow segment a startup sets out to win first before expanding. A good one is made up of customers who buy for similar reasons, buy in a similar way and talk to each other, so each customer you win makes the next one easier.
Can a startup have more than one ICP?
Eventually, yes. At the start, one is far more effective, because your message, product and channel can all be built for the same customer. Add a second profile when the first segment is producing customers predictably and you have evidence that a neighbouring group has the same problem.
Where Gufy fits
Gufy helps SaaS and mobile-app startups decide who to sell to first and prove it: go-to-market strategy to choose the segment and the message, then landing pages, paid campaigns and tracking to test whether that segment responds. A growth workshop is often the quickest way to get from "everyone" to one profile the whole team agrees on.
If you are about to spend on marketing and cannot yet name your first hundred customers, start there.