How to choose your ideal customer profile and first B2B segment

An ideal customer profile (ICP) is a short description of the kind of company that gets the most value from what you sell and is most likely to buy it soon. You find yours by looking at your best customers and conversations so far, then writing down what they share in terms you can search for. From there you narrow it to one first segment: a group of companies small enough to reach directly and specific enough that one message fits all of them.

The goal is not a perfect definition. It is a working guess you test with a small batch of outreach, then keep or change. Founders who skip this step often write to "small businesses" in general and cannot tell why nothing comes back.

What an ICP is, and what it is not

Think of the ICP as a description of companies, not people. A buyer persona is the next level down: the role inside that company who feels the problem and can say yes, such as a head of operations or the founder. You need both, but the company comes first: it decides whether the problem exists at all.

Compared with your total market, the ICP is narrow: the part where the need is sharp today, not everyone who could use the product. For a founder doing their own outbound (contacting people who have not reached out first), a broad target is expensive, because every message has to stay generic.

Advice on B2B lead generation often starts with channels and tools. A small team tends to do better the other way round: decide who to reach, then how.

Three layers: firmographics, pains and buying triggers

Build the profile in three layers, each answering a different question. A common mistake is a profile that stops at the first layer.

Firmographics

Firmographics are company-level facts you can filter on: industry, headcount, location, revenue band, business model, the software they already use. They matter because they are searchable: a profile you cannot turn into a list is one you cannot reach.

Industry and size alone do not predict buying. Two companies that look identical on paper can have very different problems.

Pains

Describe the pain, the specific problem your product removes, in the words a customer would use, ideally words you have heard on a call. "Inefficient workflows" is not a pain. "We find out a client is unhappy only when they cancel" is a pain.

To check it, ask whether someone in the target role could read your pain statement and say "that is us" or "not us" within seconds. If the honest answer is "it depends", it is too vague.

Buying triggers

Buying triggers are events that make the pain urgent now rather than someday. Common examples:

• a new hire in a role that owns the problem
• a funding round or a new budget cycle
• an expansion into a new country
• a decision to drop a tool they used before

Many companies that fit your ICP are not looking for a solution this month. A trigger narrows the list to those more likely to care today. Some, like a job post, are visible from outside; others you learn only by asking.

Start from the customers you already have

Your first customers hold the pattern, even if there are only a handful. List every paying customer, serious pilot and prospect who had a real conversation with you. Include the ones that went badly. For each, write down:

• the company, in firmographic terms
• who brought you in and who signed
• the problem they described, in their own words
• what happened just before they started looking
• how long the decision took, and whether money was a hard topic

Sort the list so the ones who got the most value and decided fastest sit at the top. Look at what the top group shares that the bottom group does not. Often it is not the industry but the business model, a stage of growth, a shared tool or the same event happening to each of them.

With no customers yet, use waitlist sign-ups or interviews you ran while building. The pattern will be weaker, so treat your first test as a learning exercise.

Pick one first segment

Choose one concrete slice of your ICP to contact first. Running several at once is tempting, but at small volume each gets too little data to read.

Write the segment as a single sentence that names the firmographics, the role you will contact, and the trigger if you have one. If you cannot write that sentence, the segment is not ready.

Urgency and budget

When you compare candidate segments, two questions do most of the work.

First, is the pain urgent? A problem that costs them money or customers this quarter beats one that is merely annoying. Good signs: they already spend time or money working around it, or have tried another solution and dropped it.

Second, can they pay? A segment that likes the idea but has no budget will give you friendly conversations and few decisions. Prefer segments where your contact controls a budget, or where the problem already has a cost to compare your price against.

Last, make sure you can find these companies and the right person inside them. Otherwise the segment is not usable yet, however good it looks on paper.

Example: finding a first segment from six customers (illustrative)

Everything here is invented for illustration: the founder, her customers and the numbers.

Picture a founder selling a tool that sends invoice reminders and matches incoming payments. She has six customers: three small marketing agencies, one software consultancy, one dental practice and one construction subcontractor. The four service firms were running within days and use it weekly. The dental practice canceled. The subcontractor took months to sign.

The four good customers do not share an industry. They share how they work. All of them bill by project milestone and have 10 to 50 people, and in each the owner still chases late payments personally. In three of the four, the trigger was a large client paying very late, which put pressure on payroll.

Her first segment becomes: owner-led service agencies in the US with 10 to 50 people that bill by project milestone, contacting the founder or managing partner. From what she heard on calls, she drafts a pain statement to test: "Chasing late invoices takes the hours I should spend selling." She keeps construction as a possible second segment and sets healthcare aside for now.

Test the segment with a small batch of outreach

Whether you run this test yourself or with outside help (outbound sales without a sales team compares the options), decide in advance what you are testing, and change one thing at a time: here, the segment. Keep the message structure and the sending setup stable, so you know what caused the result. For the emails themselves, see writing a short B2B cold email sequence.

Before the first send, check the cold outreach rules in the country you target: is B2B cold email legal? covers the US, the EU and the UK. Also check that your emails reach inboxes, since a quiet campaign can mean a wrong segment or messages landing in spam. See cold email deliverability basics.

Batch size is a judgment call. A common practice is 100 to a few hundred contacts per segment, small enough not to use up a narrow market on a first draft. Treat it as a rough habit, not a rule. Such a batch often gives only a few replies to read, so take them as clues, not proof.

Read the replies, not only the count, and sort them into groups:

• "Yes, this is a problem for us." The pain is real in this segment.
• "Not right now." The fit may be right, but the trigger is missing.
• "Talk to someone else." The segment may be right, but you reached the wrong role.
• "We don't have this problem." The segment or the pain statement is wrong.

Note the words people use. If they describe the problem differently, borrow their phrasing for a later batch to the same segment, never in a batch that also switches segments, or you test two things at once.

When to switch, and when to stay

Switch segments when "we don't have this problem" keeps coming from people who clearly understood your message. That suggests the pain is not there.

Stay and adjust when the answers point to something smaller. Many "not right now" replies suggest adding a trigger filter. Many "talk to someone else" replies suggest a different role. Silence is the hardest to read: check deliverability and the clarity of your message before blaming the segment.

When you do switch, take the next segment on your shortlist and keep the same message structure. Only the structure carries over: the pain wording and the role change with the segment by design, so the comparison stays rough. Keep a short log per segment of who you contacted, what you said and what came back. After a few segments, that log tends to teach you more than any brainstorm.

Key takeaways

• An ideal customer profile describes the kind of company that gets the most value from what you sell and is most likely to buy it soon. A persona describes the person inside it.
• Build the profile in layers: searchable firmographics, a pain in the customer's own words, and a trigger that makes it urgent.
• Look for the pattern in your first customers, including the ones that went badly.
• Pick one first segment, written as a single sentence, and weigh urgency and budget first.
• Test it with a small batch, keep the message structure and setup stable, and read the replies rather than only counting them.
• Switch when people who understood you say the problem is not theirs. For smaller signals, adjust the trigger or the role instead.