In 2019 I posted a short list of questions I ask myself. The first one was: "What are the three primary problems my target market has? How sure am I that I know this?"

Seven years and one market research company later, the second sentence is still the one that matters. Almost every B2B company can describe its ideal customer. Very few can say how sure they are.

What is an ideal customer profile (ICP)?

An ideal customer profile (ICP) is a description of the companies most likely to buy from you, get value quickly, stay and grow. A useful ICP says who they are, what situation pushes them to buy, and why they care about what you do better than anyone else. In B2B it describes the company, not the person. The people inside it are your buyer personas.

An ICP is a model of a buyer, not an actual buyer. It's a hypothesis. That's the part most guides skip: they show you how to write one, never how to check if it's true.

This guide covers both. Here's the short version:

ICP vs buyer persona vs target market

These three get mixed up constantly, and it matters, because each one answers a different question.

Your ICP tells you where to look. Your buyer personas tell you who to talk to once you're there. You need both, in that order.

Why most ICPs are wrong

Most ICPs fail in one of four ways.

1. They're only firmographics. "B2B SaaS, 200 to 2,000 employees, North America." That's not an ICP, that's a filter in a data tool. Your three closest competitors have the exact same one. Firmographics tell you who could buy. They say nothing about who needs to buy now, or why they'd pick you.

2. They're built only from current customers. Customers are the people who already said yes, despite your messaging, your pricing and your missing features. Talking only to customers is survivorship bias. They can tell you why you won. They can't tell you why the rest of your market never shortlisted you.

3. They're built from the biggest logos, not the best customers. The logo on the homepage is often not the customer who renews, expands and refers. Big is not the same as best.

4. Nobody checked. In Wynter's How B2B Marketing Leaders Make Decisions study, 53% of B2B SaaS marketing leaders said they research non-customers once a year or less (n=100). When we asked what their last big marketing decision was based on, 32% described a gut or leadership call. Only 9% named direct research with target buyers.

One AVP of Product Marketing summed it up: "Asked sales what they think, we argued about our opinions, and no one bothered to ask anyone in our target audience."

A wrong ICP is expensive, and it hides for a long time. On my podcast, Guy Cohen, CEO of Wonder, told me how they grew to about $8M in revenue before realizing their ICP was wrong, then tore it down and rebuilt it. Revenue can grow on the wrong customers for years. It just stops compounding.

How to define and validate your ICP in 5 steps

1. Find your best customers in your revenue data.
2. Find out what situation made them buy.
3. Pick the three attributes that actually predict fit.
4. Validate the profile with your target market, not your customer list.
5. Write it on one page, with disqualifiers.

Steps 1 to 3 build the hypothesis. Step 4 tests it. Most companies stop after step 3.

Step 1: Find your best customers (not your biggest)

Start with data you already have. Export every customer from your CRM and billing system with these columns, then sort.

Then cut revenue by segment: company size, industry, use case, acquisition channel. Look at share of revenue and share of customers side by side. The gap between the two is where your ICP hides.

Here's what that looked like when I did it for Wynter. Companies with 1,000+ employees were 18% of our customers but 41% of our revenue. The 51 to 200 employee segment had the most customers of any size band, but brought in about 20% of revenue.

The numbers weren't the real lesson, though. When I looked at which accounts stayed, the pattern wasn't size. It was need. Those who come in with a one-off project don't make good long-term customers. A best-fit customer is somebody with many ongoing needs: a product launch, then new positioning, then a brand baseline. Company size was just a proxy for having that many needs.

This pattern isn't unique to us. In ChartMogul's SaaS Retention Report (data from 2,100+ SaaS businesses), top-quartile companies with average revenue per account over $500 a month had 109.3% net revenue retention. For accounts under $10 a month, top quartile was 65.1%. Who you sell to sets your retention ceiling before your product does.

Two warnings:

• Small numbers lie. If you have fewer than 30 customers, don't calculate percentages. Read every account and look for the story.
• Revenue data only shows who bought. It can't show the better segment you never reached. That's what step 4 is for.

Step 2: Find out what situation made them buy

Firmographics describe a company. A situation describes a moment. Companies don't buy because they have 500 employees. They buy because something happened: a new CMO, a failed launch, a competitor repositioning, a board asking for proof.

Interview 8 to 12 of your best customers from step 1, ideally ones who bought in the last 6 to 12 months while they still remember. Ask about the past, not the future:

• What was happening in the business when you started looking?
• What were you doing about the problem before? What did that cost?
• Who else was involved in the decision, and what did each of them care about?
• What else did you consider, including doing nothing?
• What almost stopped you from buying?
• What would you tell a peer about why you picked us?

Jenni Romaniuk at the Ehrenberg-Bass Institute calls these moments buying situations, or category entry points. Her report for the LinkedIn B2B Institute, Category Entry Points in a B2B World (2022), gives a useful checklist: when, where, while, with what, why, how feeling, and with or for whom. "To help launch a new product to our customers" is a buying situation. "To integrate easily with our financial reporting software" is another. Each one is a slice of your ICP you can target and message.

Superhuman did a version of this with a single survey. Rahul Vohra asked users how they'd feel if they could no longer use the product. Only 22% said "very disappointed", well under Sean Ellis's 40% benchmark. Then he looked at who those 22% were: founders, managers, executives and business development people. Counting only that segment, the score rose to 33%. He wrote a profile of that person, "Nicole," a busy professional who reads 100 to 200 emails a day, and built for her.

The lesson: don't average your customers. Find the segment that would be very disappointed to lose you, and study them.

Step 3: Pick the three attributes that actually predict fit

An ICP with 15 attributes is a wish list. Lenny Rachitsky asked around 30 B2B founders, from Gong, Gusto, Looker and others, how they found their first ICP, and every founder landed on at least three attributes. Gong's first ICP was US, English-speaking software companies that sold over video calls with deal sizes of roughly $1K to $100K. That was about 5,000 companies.

Choose your three from the attributes that separated best customers from the rest in steps 1 and 2:

April Dunford's test is the best filter I know: your best-fit customers are the ones who "really care a lot about your unique value." Plenty of companies care a little. Your ICP is the ones who care a lot, and the attributes are whatever makes them care.

If an attribute doesn't change who you'd target, what you'd say, or what you'd build, cut it.

Step 4: Validate your ICP with your target market

This is the step almost nobody does, and it's the one that makes the difference.

Everything so far came from customers. But up to 95% of B2B buyers aren't in market at any given time (John Dawes, Ehrenberg-Bass Institute, 2021). And when they do buy, they mostly pick from a list they already had. In a Bain and Google survey of 1,208 US buyers, 90% chose a vendor that was on their short list at the start of the process. Your ICP has to describe the people who will buy next, not only the ones who bought last.

So take your hypothesis to the market. Survey people who match the profile, most of whom have never heard of you, and check five things:

Ask open questions first and multiple-choice second. If you show options first, people just pick yours.

The real power is comparing segments. Run the same survey with two or three candidate ICPs, say mid-market SaaS vs enterprise vs fintech, and compare which one ranks your problem highest, has the clearest trigger and the most reachable budget owner. That turns an internal argument into a decision. In the same Wynter study, 31% of marketing leaders said research changed their targeting, ICP or segments in the past year.

How many people you need:

Don't forget the buying group. Gartner says a complex B2B purchase involves six to ten decision makers. In Wynter's survey of how B2B SaaS CMOs buy software (n=101), the average buying committee was 5.1 people and 62% evaluated exactly three vendors. Survey more than the person who signs. In most cases it's the manager-level folks who actually build the shortlist.

If you'd rather not build the panel yourself, this is exactly what a Wynter target customer survey does: verified B2B professionals matched to your ICP by job title, seniority, company size and industry, with answers in 12 to 48 hours.

Step 5: Write your ICP on one page (template)

Put the ICP on one page that sales, marketing, product and customer success all use. Here's the template, filled in with Wynter's own ICP so you can see what specific looks like.

The disqualifiers row is the one most ICPs leave out, and the one sales will thank you for. Note that our fit depends on who our customers sell to, not what they do. If their buyers aren't in our panel, we're the wrong choice no matter how good the company looks on paper.

Notice also what's not on it: age, gender, hobbies. That belongs to consumer marketing.

Ideal customer profile examples

Sources: Lenny's Newsletter for Gong, Gusto and Looker; First Round Review for Superhuman.

How often should you update your ICP?

Review it once a year, and re-check it whenever something big changes. Pain points and buying processes shift slowly. Triggers and how buyers see you vs competitors move faster.

AI is the current example. In our decision-making study, the question marketing leaders most wanted answered was how AI is changing their buyers (35%). If your ICP research is from before your buyers started using AI to shortlist vendors, it's out of date.

When ICP research is the wrong move

• You have fewer than 10 customers. Talk to every prospect yourself. Founder calls beat any survey at this stage.
• Your whole market is a few hundred named accounts. No survey panel will reach enough of them. Interview them directly.
• Your buyers are outside the panel's strength. Every panel has strong and weak segments. Wynter is strongest in B2B tech roles and weak outside tech. Check feasibility before you buy anything.
• The decision is already made. If leadership won't change course whatever the data says, research is theater. Save the budget.

Common ICP mistakes

• Too many ICPs too early. Three ICPs means three sets of messaging, three content plans and a sales team that can't remember any of them. Nail one, then add the next.
• Confusing TAM with ICP. Your total addressable market is everyone who could buy. Your ICP is who you go after first.
• Letting one team own it. If sales has an ICP and marketing has another, you have none.
• Treating intent data as an ICP. Intent data tells you who's researching a topic this week. It doesn't tell you who's a good customer.
• No evidence column. If nobody can say where a line in the ICP came from, it's an opinion.

FAQ

What is an ICP in marketing and sales? An ideal customer profile (ICP) describes the companies that get the most value from your product and are most likely to buy, stay and grow. Marketing uses it to decide who to target and what to say. Sales uses it to qualify accounts and prioritize the pipeline.

What's the difference between an ICP and a buyer persona? The ICP describes the company: size, industry, situation and needs. A buyer persona describes a person inside that company: their role, goals, pains and how they buy. Define the ICP first, then build personas for the people in its buying group.

How many ICPs should a B2B company have? One to start, two or three at most. Each ICP needs its own messaging, channels and proof, so more ICPs means less focus. Add a new one only after the first is working.

How do you validate an ideal customer profile? Check it against two sources. First, your revenue data: do companies matching the profile retain, expand and close faster? Second, your target market: survey people matching the profile, including non-customers, to confirm they rank your problem highly, have a buying trigger and can reach a budget owner. Compare two or three candidate segments with the same survey.

What should an ICP template include? A one-line ICP, company attributes, the buying situation or trigger, top pains in buyers' words, the buying group, why they pick you, observable fit signals, disqualifiers and the evidence behind each line.

How often should you update your ICP? Review it once a year and re-check it whenever you enter a new segment, launch a new product, see retention or win rates drop, or a competitor repositions.

How do you create an ICP without customers? Use the market instead of your customer list. Interview 15 to 20 people in two or three candidate segments, then survey 100 per segment to see which one ranks the problem highest and has a clear trigger. Treat the result as a starting hypothesis and revisit it once you have 20 to 30 customers.

Related guides

B2B buyer personas · Voice of the customer for B2B · B2B market research · Competitive analysis template · Brand perception survey · The Wynter Research Playbook

Sources

• Wynter, How B2B Marketing Leaders Make Decisions, October 2026 (n=100)
• Wynter, How B2B SaaS CMOs Buy Software, February 2026 (n=101)
• Lenny Rachitsky, How to identify your ideal customer profile (ICP), Lenny's Newsletter, 2023
• First Round Review, How Superhuman Built an Engine to Find Product/Market Fit
• April Dunford, A Quickstart Guide to Positioning
• Jenni Romaniuk, Category Entry Points in a B2B World, The B2B Institute at LinkedIn, 2022
• John Dawes, Advertising effectiveness and the 95-5 rule, Ehrenberg-Bass Institute, 2021
• Harvard Business Review, What B2Bs Need to Know About Their Buyers, 2022 (Bain and Google survey, n=1,208)
• Gartner, The B2B Buying Journey
• ChartMogul, SaaS Retention Report (2,100+ SaaS businesses)
• How to Win podcast, Rebuilding the ICP with Wonder's Guy Cohen