The Founder's ICP Template: One Page, Five Questions, Twenty Minutes

Download any ideal customer profile template and you get the same thing: thirty fields, most of them company size, industry, revenue band, headcount, tech stack.

Ask who wrote it. Almost always a data company, and those thirty fields are the exact filters their database sells. The template is not advice. It is the top of somebody else's funnel, and filling it in teaches you to describe your customer in the only language a bought list can read.

You have no customers and an afternoon. You need a page you can fill in today and use tomorrow to find five real people. That page has about five questions on it, all of them open, none of them a dropdown.

Here it is, with a worked example that was wrong before it was right.

Quick answer

A founder's ICP template is one page with five open questions on it: where this person already spends time, how they talk about the problem, what they are visibly frustrated by right now, what they are like as a person, and what they have actually done about it. No headcount, no revenue band, no tech stack. Those are filters for buying a list, and you are not buying a list. You can fill this in twenty minutes with zero customers, and the test of a good answer is that you could recognise the person in a comment thread. MyPip builds a first version from your website overnight, but the page is more useful when you have argued with it yourself.

What actually goes on a founder's ICP template?

Five questions, written as prose, not as fields. Copy these into a document and answer each one in a couple of sentences. If an answer could describe a thousand companies, it is not an answer yet.

1. Where do they already spend their time? Name rooms, not categories. Not "online" or "LinkedIn", but the specific subreddit, the Discord, the Slack group, the comments under a certain kind of post. If you cannot name the room, you cannot show up in it.

2. How do they describe the problem in their own words? Their language, not yours. You might sell "customer research". They wrote "I built the thing and I have no idea who to message." Write down the sentence they would type into a search bar. That sentence is the raw material for every message you send later.

3. What are they visibly frustrated by this week? Something said out loud, in a post, a comment, a bio. A live frustration is a raised hand. A demographic is not.

4. What are they like as a person? The soft correlates that no form has a field for. How they think, what they value, what they are proud of, what they are quietly worried about. This feels unscientific and it is the field that predicts the most.

5. What have they done, rather than who are they on paper? Have they posted about this. Have they commented on someone else who did. Is the problem in their bio. Behaviour qualifies; a job title is a tie-breaker at best.

That is the whole template. Five open questions instead of thirty dropdowns, because you are profiling a person you would recognise, not assembling a query. If you want the argument for why these five and not the other twenty-five, that sits on the reasoning behind the five fields. If you are not yet sure what the document even is, start with what an ICP is in plain English.

How do you fill it in when you have no customers yet?

From three places, in this order, and it takes about twenty minutes.

Yourself, if you built this for a problem you had. You are one data point and you are a real one. Write down what was true of you when the problem bit. Where were you, what had you tried, what did you type into Google at eleven at night.

The rooms, for the language. Go and read fifty threads where people describe this problem. Do not post anything. Copy the phrasings that keep repeating, especially the ugly ones. That is question two answered in their words rather than yours, and it is the single highest-value twenty minutes on this list.

The people who already reacted, if any exist. Anyone who replied, signed up, asked a question, or said "oh that's interesting" at a party. Even five of them. Look for what they have in common that is not on their business card.

Then write it down and commit to it, knowing it is wrong. A first ICP is a starting block, not a finished thing. Michelangelo started with a block of marble and chiselled until the statue came out. Your first page is the block. The chiselling is what happens when real people start replying.

What does a filled-in ICP look like?

Here is mine, in the wrong version and then the right one, because the wrong version is more instructive.

The version I wrote first. Sales reps. SDRs with a quota to hit, because a tool that finds the right people is obviously a better way to hit a number. Hard fields, mostly: job title, team, target. It read like a proper ICP.

It was wrong in a way the fields could never show. People with a quota want maximum output. Send the most messages, book the most meetings, and quality is somebody else's problem. That is a brutally competitive market and it is the opposite of what I built. Nothing in "SDR, mid-market, quota-carrying" told me that. The trait did.

The version the data wrote. The people actually saying yes were not chasing quotas at all. They were founders and early marketing hires who had built something and had no natural way to talk about it. That is a different person with a different frustration, and finding it changed the product, not just the messaging.

And a second one, from before MyPip. At Plastic Bank the thesis was a sustainability or marketing director, company of 250 to 500 people, this revenue, that product volume. Then we did a deep dive on who had actually bought. The correlation with that thesis was essentially zero. What did correlate was personality: avid lovers of the ocean, humble people, often a bit older, often having lived in a few countries and worked remotely. Not one of those is a field in any template you can download, and all of them are what let us go and find more of the same people.

Both examples fail in the same direction. The hard fields were confidently wrong and the soft ones were quietly right.

Is a B2B ICP template different?

Slightly, and less than the enterprise versions want you to believe. In B2B you are profiling two things at once: the company that pays and the person who cares. Firmographics earn a small role in the first, mostly as a disqualifier, and they are useless for the second.

The practical version at your size: keep the same five questions aimed squarely at the human, then add one line about the kind of company they work at, written as a constraint rather than a target. "Small enough that this person can just decide" is a more useful sentence than "50 to 200 employees, Series A". One tells you whether a conversation can go anywhere. The other tells you what to type into a database.

The reason this matters more than it sounds: matching your description does not mean ready to buy. Most people who fit your profile perfectly are not in the market this month, which is the small share of them who can buy this month and why the frustration question does more work than the firmographic one.

How narrow should each answer be?

Narrower than feels sensible. Not SaaS founders. SaaS founders building healthcare products, doing between zero and a million in revenue, based in lower Montreal. That level of specificity feels absurd when you write it, and it is the point at which the page becomes usable.

The fear is that narrowing shrinks the market. It does the opposite. Jason Cohen's "Speaking to Carol" describes it as concentric rings: write for one sharply drawn person and you also reach the ring of people who are nearly her, roughly ten times bigger, and then a broader ring perhaps a hundred times bigger where a fraction still choose you. Subaru marketed deliberately to one small, specific segment in the 1990s and then grew faster than any other carmaker for a decade, mostly selling to people outside it.

The mechanism is recognition. Someone who matches three of your five answers still sees themselves in a message written for all five, and does the translating themselves. A message written for everyone gives nobody anything to recognise.

What do you do with the page once it is filled?

Turn it into a search, the same day. That is the entire point of the document, and the reason it is written in soft language rather than filter language: every answer on the page is something you can go and look for in public.

Question one gives you the rooms to open. Question two gives you the exact words to search for. Question three tells you which of the results are worth your five minutes, because a person who complained about this on Tuesday is a different proposition from a person who merely fits the description. Then you turn the page into a search and start with one person, not a list.

The message you send is built from the same page: their words from question two, their live frustration from question three, and nothing at all from a firmographic field. That is the message that comes out of it.

When should you rewrite it?

When conversations start dragging. The tell that your profile is wrong is not a low conversion rate, it is speed. If you are writing long explanations, if they are asking a pile of questions, if they cannot see what it is worth to them, if it takes a dozen messages to get anywhere, you have the wrong person or the wrong words for them. With the right person, everything is quick and simple. They arrive already understanding what you do.

I have seen the fast version once and never forgotten it. At Baseline Media a client got on a call opening with "that's a service I want, and I know you can do it." No convincing, no explaining. That speed was the proof we had profiled them correctly, long before any revenue confirmed it.

So keep chiselling. Let the people who reply overwrite the page, not the other way round. Every reply, every conversion, every fast yes is data about who this is really for, and the page should look different in three months or you have not been reading your own inbox.

If you would rather not start from a blank document, paste your site and I will take a first pass overnight: the profile pulled from what you have already written, then a first draft built from your site overnight that you can argue with. Arguing with a draft is faster than staring at an empty page, and the arguing is where the real profile shows up.