How to create an ideal customer profile

You have a product and one or two customers, and every piece of advice starts with "define your ideal customer profile" first. Most templates for how to create an ideal customer profile ask for a sales team, a CRM and a hundred customers you do not have, so you guess, and your messages, your homepage and your postssoon all speak to nobody in particular. That is one of the most common reasons no one is buying. Here you write two or three short guesses about who your best customer is, test them on real people at volume, and keep the one that gets the most yeses.

Do these three things

About 2 hours today, then about 4 hours a day. Most founders can pick one ideal customer after about 15 days.

Show the three actions in fullHide the full list

1. 1Today: list everyone who has paid, tried or talked to you, and write 2 or 3 one-sentence guesses of your ideal customer: who they are, their size, the moment the problem hurts, what they use today, and who it is not for.
2. 2Every day: send 100 personal messagessoon to people who fit, split across your guesses, and ask for a 15-minute chat about what made them look for a fix. AI drafts each one; you add one true line about the person.
3. 3Every Sunday: count chats and yeses for each guess. After 1,500 messages, keep the guess with the most, and rewrite its sentence in your buyers' own words.

Get help doing these

ChatGPT or Claude reads this guide and works through it with you.

Do this with ChatGPT(opens in a new tab)Do this with Claude(opens in a new tab)

1. List everyone who has paid, tried or talked to you#

Your ideal customer profile starts with the people who already said yes, even if there are only two. Put every customer, trial user, reply and warm lead in one sheet, then mark the two you would most like ten more of.

With so few customers, one row can mislead you. So count everyone who showed real interest: people who paid, started a free trialsoon, replied to a message, asked what you charge after reading your pricing pagesoon, or booked a call. If you already have users, ask them for feedbacksoon with the same columns in mind. The column that matters most is "what happened just before". One founder on Hacker News had one pilot customer and could not find a second, in June 2026. The best reply told them to write down what caused the first customer's pain, because "your ICP becomes 'companies that match the trigger,' which is often not the same as 'companies in that vertical.'"

Name | Paid, trial, reply or chat | Kind of business (or person) | Size | What happened just before they needed you | What they used before | How they found you | Days from first contact to yes | What they pay | Want 10 more? (yes/no)

If you have call notes or emails, paste them into ChatGPT or Claudesoon and ask it to fill these columns. Check every cell yourself: it will guess where the notes are silent. Then mark your best two rows: the ones that paid fastest, use you most, and were easiest to work with.

• Tool: Google Sheets and ChatGPT or Claude. Both free to start.
• Time: 45 minutes.
• You will have: a sheet with one row per person, every column filled, and your best 2 marked.
• It worked if: you have at least 5 rows. With fewer, add everyone who replied, asked a price or visited your pricing page and wrote to you.
• Common mistake: starting from your own background ("I was a designer, so designers") or from friends who bought to be kind. Friends and former colleagues are polite, and polite is not the same as buying. Mark the rows by what people did, not by who you like.

2. Write your ideal customer profile as two or three one-sentence guesses#

Write one sentence per guess: who the customer is, how big, the moment the problem hurts, what they use today, and who you are not for. Start from your two best rows, and add one or two near neighbours, so you have something to compare.

An ideal customer profile, or ICP, describes the kind of business, or kind of person, that gets the most from what you sell and pays for it. A buyer persona is different: it describes the one person inside that business who decides. For now, write the ICP. You will learn the decision-maker in your chats in step 4.

[Your business] is for [kind of business or person] with [size], who [the moment the problem hurts], and today they [what they use or do instead]. They can pay [your price].Not for: [who you will turn down, and why].

Keep each guess narrow. Lenny Rachitsky's advice, from studying how Gusto, Gong and Snyk started, is to pick three very specific traits, such as size, the tool they use, or where they are, rather than a broad market. Your second and third guesses change one thing: the same moment in a different kind of business, or the same business at a different size. Never write more than three. Each extra guess cuts the messages each one gets in step 4.

Then check each guess with four questions. Cut or change any guess that fails one:

1. Does it hurt now? Something just happened that makes them look for a fix this month.
2. Can they pay your price? Check it against your price, not your hopes.
3. Can you find them by name? You will test this in step 3.
4. Do they gather somewhere? A forum, a subreddit, a trade showsoon, a Facebook groupsoon.

The "Not for" line matters as much as the rest. A founder who writes it stops saying yes to work that costs more to win than it pays, which is what customer acquisition costsoon measures, and a stranger reading your site can tell straight away whether it is for them.

If you run a D2C store: your customer is a person, not a company. Write the moment in their life ("just moved to a hard-water city", "a new baby and no time") instead of a company size. In Shopify admin, open Customers, then Segments, then Create segment, and filter by number_of_orders or amount_spent to see who bought twicesoon and where they live.

If you run an agency: your ideal customer profile is your nichesoon: one kind of client with one problem you fix. It decides who you message for your first client. "Small businesses that need marketing" is not a guess. "B2B software companies of 2 to 15 people where the founder still writes every post" is.

If you run a SaaS: name the tool they use today and the team size. It is the fastest way to find them, and it tells you which alternatives pagessoon to write later.

• Tool: a Google Doc.
• Time: 45 minutes.
• You will have: 2 or 3 sentences and one "Not for" list.
• It worked if: you show each sentence to a friend, and they can name 3 real businesses (or people) that fit it.
• Common mistake: "small businesses", "founders" or "women 25 to 45". Nobody can find those by name, and nobody reading them thinks "that's me".

3. Find 100 real names for each guess#

Before you test a guess, prove you can reach it: find 100 real businesses or people for each one, by name, with a way to contact them. A guess you cannot fill is a guess you cannot sell to.

Build a lead listsoon in the same sheet, with one extra column for the guess each name belongs to. Search where each guess actually shows up. A local business shows up on Google Maps. A software team shows up on LinkedIn, GitHub or in the reviews of a competitor. A shop brand shows up on the stockist page of the stores it sells insoon. Give ChatGPT or Claude one guess at a time and ask for 30 businesses that fit, with where it found each one. Open each one yourself and keep only the real ones.

Then find the person. For a business, that is usually the owner or the person who would sign: find their emailsoon from the company site or LinkedIn.

LinkedIn limits how much a free account can search. When your searches look like prospecting, it shows a "commercial use limit", resets it on the 1st of each month, and will not tell you how many you have left. So do not do all your searching there. Spread it across Google, Google Maps, stockist pages, trade lists and online communities where your buyers talksoon.

Step 4 uses 100 names a day, so after the first 300, add 100 new names every day, split across the guesses that are still in the test.

• Tool: Google Sheets, Google, Google Maps, LinkedIn, and ChatGPT or Claude.
• Time: about 2 hours for the first 300 names, then 45 minutes a day.
• You will have: 300 rows, each with a name, a business, a way to reach them, and a guess number.
• It worked if: you fill 100 names for a guess in about an hour. If a guess stops below 50 after an hour, widen it by one trait (a bigger size range, or a nearby region) or swap it for another near neighbour.
• Common mistake: a guess that no search tool can find, such as "brands that care about quality". Every trait in the sentence should be something you can see from outside: a size, a shop listing, a tool, a place, a job title.

4. Message 100 people a day and ask what made them look for a fix#

Send 100 personal messages every day, split evenly across your guesses, and ask each person for a 15-minute chat about the problem, not for a sale. How each guess answers is your test.

100 a day is the point. At a pass mark of 2 chats per 100 messages, 10 a day would give each guess one chat a fortnight, and you would still be guessing in three months. 100 a day gives you about 2 chats a day, and after 1,500 messages, about 15 days, each guess has had 500 tries.

Let AI do the drafting. Paste 10 names at a time into ChatGPT or Claude, each with one real fact about the person, plus the template below. Read each draft, make the first line true and about them, then send it. That is about a minute a message. Log the guess, the date and any reply in your sheet.

Send each message one at a time from your own work email, or to your LinkedIn connections. Gmail requires every sender to have SPF or DKIM, two settings that prove an email comes from your domain, and wants spam reports below 0.3%. At 100 a day, set up a second domainsoon for these messages and warm it up first, so your main email stays safe. In the US, the CAN-SPAM Act covers business email too: every message needs your postal address and a way to say no, and you must honour a "no" within 10 business days. The UK and EU have their own, stricter rules, so check them before you write to buyers there. On LinkedIn, hitting the invitation limit usually blocks invitations for a week, so message people you are already connected tosoon and send the rest by email.

Subject: [the moment it hurts, in 3–5 words]Hi [first name],[One true line about them: something they launched, won, wrote or said.]I run [your business]. I'm trying to learn how [kind of business] handle [the problem] when [the moment it hurts].Could I ask you 4 questions on a 15-minute call this week? No pitch.[Your name]
[Your business, postal address]
Not for you? Reply "no" and I won't write again.

Follow upsoon twice with everyone who has not replied, 3 days apart, with one new line each time. Stop the moment someone says no.

On the call, ask these four customer interview questions, and write down the answers in their exact words:

1. What happened just before you started looking for a fix?
2. What do you use or do today?
3. What did it cost you last time it went wrong?
4. If this were fixed, what would it be worth to you?

Their words matter more than their yes. The phrase they use for the problem becomes your value proposition and the first line of your next messages.

If you run a D2C store: email every past buyer these questions, with a one-question survey after checkout: "What made you buy today?" Fairing's free plan covers stores with up to 200 orders. Send your 100 a day as one-to-one replies to people who have posted about the problem, spread across several places, and follow each place's rules on self-promotion. On Reddit, read the rules firstsoon. The same chats are how your first sales from strangers start.

• Tool: Gmail or LinkedIn, ChatGPT or Claude to draft, and your Google Sheet. A CRMsoon can wait.
• Time: about 1 hour 40 minutes a day for messages, 45 minutes for new names (step 3), and about 1 hour for chats.
• You will have: 700 messages a week, each tagged with its guess, and notes from every chat.
• It worked if: you get at least 2 chats per 100 messages across all guesses. If you get fewer than 1 per 100 after 300 messages, rewrite the first linesoon so it is more about them, then send the next 100.
• Common mistake: deciding after 20 messages, or pitching on the call. You are testing who has the problem, so let them talk. Offer to show what you sell only at the end, and only if they ask. The discovery callsoon comes later.

5. Count results by guess every Sunday and keep one ideal customer profile#

Every Sunday, fill in one row per guess. After 1,500 messages, keep the guess with the most chats and yeses per 100 messages, drop the rest, and rewrite its sentence in your buyers' own words.

Guess | Messages sent | Replies | Chats | Said "this hurts now" | Asked for a price, trial or sample | Paid | Price they paid

The decision rule after 1,500 messages, 500 per guess:

• Keep the guess with the most chats per 100 messages. If two are close, keep the one with more people asking for a price, a trialsoon or a sample.
• Drop any guess below 1 chat per 100 messages.
• Run one more round if no guess reaches 2 chats per 100: write a new neighbour of your best guess and test it for another 500 messages. If three rounds find nothing, check the problem is real before you market it.

Then rewrite the winning sentence with the words people used on the calls, and add to the "Not for" list anyone who wasted a call. Put the sentence at the top of a one-page doc. Everything else now starts from it: your homepage headlinesoon, your listsoon, the channels you choose and your positioning. It also helps AI tools: in G2's March 2026 survey of 1,076 software buyers, 51% said they now start their research in an AI chatbot rather than Google. A buyer tells the chatbot who they are and what just happened, and a site that says the same words gives ChatGPT something to matchsoon when it recommends a business.

Ask every customer who fits the sentence for one introduction to someone with the same problem. The Hacker News reply in step 1 gave the same advice. Later you can turn it into a referral programsoon.

Once a month, check the sentence against your newest customers, and rewrite it after every 10 new ones.

• Tool: Google Sheets.
• Time: 30 minutes each Sunday.
• You will have: one sentence, one "Not for" list, and the numbers that chose them.
• It worked if: at least 7 of your next 10 new customers fit the sentence. If fewer do, the sentence describes who you hoped for, not who buys: rewrite it from the rows in step 1.
• Common mistake: keeping all three guesses "to be safe". Then every message, page and post is still written for three kinds of buyer, and none of them hears it as theirs.

Where Scout7 helps#

Scout7 speeds up steps 2, 3 and 5. It does not message anyone or run the chats: steps 1 and 4 stay yours.

Steps 2 and 5. Scout7's audience research reads public social conversations in your category and builds audience segments with pain points and quotes in buyers' own words. Use them to write your first guesses and to check the words in your final sentence. It does not replace the chats in step 4.

Step 3. Join Conversations searches social platforms for live discussions about your problem. That shows you where each guess talks, so you know where to look for names.

Step 5. Scout7 keeps your brand profile, including who you sell to, as facts you can edit. Put your final sentence there, and every later Scout7 job, from articles to posts, is written for that customer.

Your ideal customer profile checklist for the next 15 days

0 of 13 done

Work through this list with an AI assistant

It drafts the work and keeps track of what is done.

Do this with ChatGPT(opens in a new tab)Do this with Claude(opens in a new tab)

Step 1 · List everyone who has paid, tried or talked to you

• List every customer, trial user, reply and warm lead in one sheet, with what happened just before each one needed you.
• Mark the 2 rows you would most like 10 more of.

Step 2 · Write your ideal customer profile as two or three guesses

• Write 2 or 3 one-sentence guesses of your ideal customer, and a Not for line.
• Check each guess: it hurts now, they can pay, you can find them by name, they gather somewhere.

Step 3 · Find 100 real names for each guess

• Find 100 real names for each guess, with a way to reach each one.
• Every dayAdd 100 new names to your list.

Step 4 · Message 100 people a day and ask what made them look

• Every daySend 100 personal messages asking for a 15-minute chat.
• Every dayFollow up with everyone who has not replied, up to twice, 3 days apart.
• Every dayHold every chat and ask the four questions.

Step 5 · Count results by guess every Sunday and keep one

• Every SundayFill in one row per guess: messages, replies, chats, hurts now, price asked, paid.
• After 1,500 messagesKeep the guess with the most chats per 100, drop the rest, and rewrite its sentence in your buyers' words.
• Ask every customer who fits the sentence to introduce you to one person with the same problem.
• Once a monthCheck your sentence against your newest customers.

Copy listDownloadReset