Your ideal customer profile: work out who to look for next

You may have one person who paid for your product because they know you, another who comes back monthly for another sign up to your free trial, a third who takes up half your week in support, and... your mom. They all count as early customers or users (OK, mom maybe not). And their actions give you very different reasons to look for more people like them.

At this stage, you're deciding who to look for again, based on the data you gathered when someone buys your product. Once you can describe that person - or that company - in a few clear details, you've got somewhere specific to do the thing that this is all about: GET MORE CUSTOMERS.

That description is your ideal customer profileA current description of the people or companies most likely to need, buy, and get a strong result from your product. Early versions may begin with informed assumptions and should change as customer behavior gives you better evidence., usually shortened to ICPShort for ideal customer profile.. Its first version is always designed to be iterated on, becoming more specific as people use the product, pay, come back, refer somebody else, have too many crazy support asks, or leave.

Start with the evidence you have

What you can write depends on what's happened so far. If you've got no users, you can still make an educated guess pre-revenue.

We start by creating your first version at zero customers and iterate from there. If you're in the first row (nobody has tried or paid), the repeated problem and what people already do give you a provisional starting point; if you've got free users, keep their product behavior and their willingness to pay as separate questions. One payment gives you a real buying event to study, while several customers let you compare without building the whole description around a singular person.

Then go to the customers who got the best result

The customer who paid first isn't automatically the customer you should pursue next. They might have just been the one with more money, or less worry about starting something new. Instead, start with results, followed by repeat useSomebody coming back to use the product again after the first try. It may show that the problem returns or that the product has become part of what they do., referrals, cost, feedback, and ease of sale; the payment signal stays separate because it proves somebody was willing to buy, while what happened after the sale tells you whether they stayed, and if you want more customers like them.

Compare the people you've spoken to or served in this order:

Results come first because a quick sale to somebody who doesn't stick, needs a different product, and leaves immediately gives you a poor direction for the next ten sales. Repeat use is how retentionThe extent to which customers continue using or paying for a product over time. For an early product, returning and repeated use are the first signs to record. first becomes visible (and retention is what we want): they came back because the problem returned, the result continued, or the product became part of what they already do. AKA the holy grail.

You don't need to give each customer a score, or a number. It can be as easy as simply noting what happened, comparing the cases, and keeping any variance visible; a tidy average can hide the difference between one customer who got a strong result and four who barely used the product.

The three details in the pattern

We want to keep an early customer profile to exactly three easy, clear details. More than that usually creates a seventeen-field description that excludes every living person, or a list of facts that sound specific without explaining why the person would care.

So instead, we look for:

1. Who they are, or the situation they're in. This could be a company type, role, life stage, location, team size, or another condition connected to the problem.
2. What they're already using, doing, or paying for. A spreadsheet, member of staff, competing product, repeated manual job, or existing budget shows how they deal with the problem now.
3. What changed recently. A new client, deadline, hire, regulation, failed process, move, diagnosis, or public request for help may explain why the problem has become urgent.

For example, “small businesses” gives you almost nowhere to start. A provisional version might be: “I'm focusing on independent agencies with five to twenty staff who already pay a bookkeeper to chase overdue invoices when late payments delay payroll.” Each detail here should be clear in its purpose: the first locates the business, the second shows current action and spend, and the third identifies the moment that may prompt a decision.

Use this sentence for your own version:

I'm focusing on [people or companies in this situation] who already [use, do, or pay for this] when [the event that makes the problem urgent].

Say it out loud. If you wouldn't use the sentence while explaining the product to a colleague or another founder, remove the formal language and try again.

Keep the user and payer separate

Confusing the person using your product with the person paying for it trips up a lot of founders when it comes to articulating their value propositionThe explanation of why a particular customer would choose your product, including the problem it helps with, the result they can expect, and why it fits their situation better than the alternatives they already have.. For example, if you have a dev tool, the purchaser may well be completely different to the users, so the words you use to describe it should change. Selling into solo devs? Go nuts: be as technical as you like. Selling to medium, large, or even small companies or enterprises? The person with the budget sign-off might sit in operations or the c-suite, and your level of technical language, and therefore ICP, should alter accordingly.

For a business product, record each part separately:

• Company or situation: the conditions that make the problem likely;
• UserThe person who uses the product or experiences the work it changes. A user doesn't always choose or pay for the product.: the person who experiences the problem or uses the product;
• PayerThe person or budget that supplies the money for a purchase. The payer and user may be different people.: the person or budget that pays for it;
• ChampionThe person inside an organization who wants the purchase to happen and helps it move through the decision process, even when they can't give final approval.: the person who wants the purchase to happen and keeps it moving;
• Final approverThe person who can give the last approval required for a purchase or contract.: the person who can give the last yes, when that isn't the payer or champion.

A self-employed buyerThe person involved in choosing or approving a purchase. The buyer may use the product, pay for it, or involve other people who do. may fill every role. In a larger company, a team member may use the product, their manager may push for it, finance may control the budget, and a senior leader may approve the contract; security, legal, or procurement may also need to check it before anybody can buy.

Each person may want a different result from the same purchase. The user may want the job to take less time, the champion may want the problem dealt with before it puts more pressure on them, the payer may need to defend the cost, and the final approver may be most concerned about making the wrong call. Keep only the answers supported by something the person said or did.

Where a difference is clear, it should change what you say and show: the user may need to see how the product works day to day, while the champion or final approver may need proof they can use to explain the purchase internally.

For a consumer product, you may only need one line for the person and another for the purchase situation. Keep them separate when one person buys for somebody else, such as a parent buying for a child or somebody choosing a gift.

Record why some customers aren't right

People who leave, decline, or struggle with the product help you see the edges of your profile, as long as you record the reason. “Poor fit” is too vague to help later; write down what happened.

Somebody may need a different product, lack the budget or authority to buy, require more support than you can provide, cost too much to find or serve, get a poor result, feel no urgency, or haggle over every part of the sale. A missing feature may be worth fixing when several otherwise strong customers need it; a customer whose problem, expected result, or route to purchase is completely different points you elsewhere.

One surprising customer doesn't require an immediate rewrite. Add the case as a question and look for another; when the same behavior appears again, you've got a reason to create a new version.

Can you find ten more?

Use your three details to find ten real people or companies, then add a source you can open for each one. You might use company pages, job posts, customer reviews, public forum discussions, professional directories, or people from your earlier validation work.

Treat this as a check on whether your description points to people who exist, whether the details are visible before you speak to them, and whether you know where to find them. It doesn't prove those ten people will buy or give you a prospecting target. If you can't find ten, work out whether the profile is too narrow, depends on private information, or describes people you can't currently reach.

Keep your rough market-size and unit economicsThe money made and spent for one customer, order, or other basic unit of the business. It usually compares revenue with the cost of winning and serving that unit. work from the Validation chapter beside this list. This chapter finds the group that appears to get the strongest result; the next chapter, Niching down, helps you decide whether the group is large enough and affordable enough to reach and serve as your starting market.

Some products need more than one profile

Some products depend on two groups, or on enough people being able to find and interact with one another. One profile can't describe both sides of that purchase, so each side needs its own record before you decide where to start.

A marketplaceA product or service that helps two or more groups exchange something, such as buyers and sellers, guests and hosts, or clients and specialists. needs a separate description for the people supplying something and the people buying or requesting it. Start with supply when demand would otherwise arrive to an empty marketplace; start with a concentrated group of buyers when suitable supply already exists and suppliers need proof that buyers will turn up. Either way, don't merge both sides into one profile.

Products with a network effectWhen a product becomes more valuable to each participant as more of the right people join and interact. also need you to record who joined, what they did, who they interacted with, and whether the other people received a result. Activity may arrive before revenue, so keep participation, repeat use, referralAn existing customer or user introducing or recommending the product to somebody else., and payment as separate events.

For enterpriseA large organization where several people and formal checks may be involved in buying a product, including budget, technical, security, legal, and procurement approval. products, add the company, user, champion, payer, final approver, and each technical or purchasing check to your record. A letter of intentA document recording that a person or organization intends to take a stated action under specified conditions. It can move an enterprise purchase forward; it isn't completed approval, use, payment, or a guaranteed sale. may show that somebody will spend time and reputation moving the purchase forward; it doesn't show use, results, completed approval, or payment.

Six customers can tell you more than ninety-four

When positioning expert April Dunford reviewed a database product she was working on, she called 100 customers. Ninety-four barely remembered buying it or weren't using it; six had changed how they ran mobile or field work because of it. Those six shared a situation and a result that the much larger group didn't, giving the company a far better set of customers to study.

Dunford had 100 customers and a team, while you may be working from one payment or three free users; your first description will carry more uncertainty. The principle still applies: begin with the people who got the strongest result, then ask what was already true about them and what made them act.

Your working customer record

The ideal customer profile record brings the comparison, three details, separate buying roles, reasons to rule somebody out, and ten real matches into one place. It also gives you space for two alternative profiles, so you can see what you rejected when you review the decision later.

Complete only the fields your current evidence supports. Blank spaces and “I don't know yet” are more accurate than details chosen because they sound plausible.

A prompt to turn your records into a working profile

Purpose: Compare the customer evidence you've collected and suggest two or three possible profiles without filling the gaps with invented people or behavior.

When to use it: Once you've completed the customer comparison in the record; you can use it with no customers, provided every conclusion is treated as an assumption.

I need to turn my customer records into a working customer profile that tells
me who to look for next.

My context
- Product or service: [what I'm offering now]
- Product stage: [idea / prototype / lightweight MVP / pre-launch / live]
- Business model: [how I expect it to make money]
- Result I promise: [the change I want the customer to experience]
- Problem I've validated: [the problem, the recognizable group, and the evidence]
- Group I'm currently considering: [my provisional customer group]
- What I've learned about market size and unit economics: [facts and assumptions]
- Validation record: [paste or attach]
- First-customer record: [paste or attach]

My customer or user cases

For each case, I'll provide:
- Name or record label
- Company or relevant situation
- Source and date
- Whether they were interviewed, used it free, paid, or declined
- Result they achieved
- Whether they returned or kept using it
- Whether they referred anybody
- Cost and time required to reach and serve them
- Feedback in their words
- How the sale or invitation went
- Who used, paid, pushed for, and approved the product
- What each person said or did that shows what they wanted from the purchase or
  were concerned could go wrong
- What they already used, did, or paid for
- What had changed recently
- Why they may be wrong for the product

[Paste each case separately. Write “unknown” where information is missing.]

Ten real people or companies I think may match

For each one, I'll provide a name or company, the three visible details, and a
source URL I've opened. If I haven't found ten, I'll leave the remaining entries
blank.

[Paste my list]

Your task
1. Review every customer or user case separately before looking for a pattern.
2. For each case, separate facts, my assumptions, your interpretation, and
   missing information.
3. Compare cases in this order: result achieved, repeat use, referral, cost to
   reach and serve, quality of feedback, and ease of sale. Keep payment on its
   own line as evidence that somebody bought.
4. Identify recurring details about who or what situation had the problem, what
   they already used or did, and what changed recently.
5. Keep the company or situation, user, payer, champion, final approver, and any
   technical or procurement checks separate. Say when one person fills several
   roles, then identify any different result or concern for each person only
   when the supplied evidence supports it. Explain what this changes about the
   words or proof I use, and write “I don't know yet” when the information is
   missing.
6. List conflicting cases and the specific reasons somebody may be wrong for
   the product.
7. If this is a marketplace, review supply and demand as separate profiles.
   Start with supply unless my evidence shows that suitable supply already
   exists and suppliers need proof of concentrated demand before joining. If
   the product relies on a network effect, keep joining, interaction, the result
   for the other side, repeat use, referral, and payment as separate events.
8. Produce two or three possible customer profiles. Each must contain exactly
   three observable details: who or situation, current action or alternative,
   and recent change.
9. Recommend one dated working profile and explain which observed actions
   support it, which assumptions remain, and what would change the choice.
10. Review only the ten possible matches I supplied. Say which appear to match,
   which don't, and what I need to check myself.
11. Recommend the next customer behavior I should watch and the point at which
    I should review the profile again.

Rules
- Don't invent or infer customer behavior, quotations, payments, results,
  repeat use, referrals, costs, market size, names, companies, source links,
  budgets, decision authority, willingness to buy, personal ambitions, or
  concerns.
- Don't create names to fill my list of ten.
- Don't treat interviews, compliments, free use, letters of intent, and payment
  as the same result.
- Don't hide contradictory cases or average unlike customers together.
- Don't assign numerical scores unless I've supplied a calculation that requires
  one.
- Use “I don't know yet” where the evidence is missing.
- Keep facts, my assumptions, your interpretation, and your recommendations
  visibly separate.

Return exactly these headings
## Review of each case
## Comparison of customer actions
## Facts
## My assumptions
## Model interpretation
## Conflicting cases and reasons to rule somebody out
## Company, user, payer, and approval roles
## Possible profile 1
## Possible profile 2
## Possible profile 3 (only if the evidence supports it)
## Recommended working profile
## Review of my ten possible matches
## Missing information
## Next behavior to watch
## Next review

Read the response beside your records, and check every pattern yourself; the model can compare what you give it, although it can't recover a missing conversation or decide that an assumption happened in real life.

Good enough for now

Your customer profile is ready for the next chapter when it contains one specific sentence with three clear details; the customer actions supporting it are visible; the user, payer, and other buying roles are separate where they need to be; you've written down why somebody may fall outside it; and you've found ten real people or companies who appear to match.

Review it after five more meaningful customer actions, or sooner when a payment, result, repeat use, referral, or departure directly challenges what you've written. One edge case should prompt a question; repeated behavior gives you a new version.

Next, you'll decide whether this is the group you want to focus on first in Niching down.

Sources

• Dunford, April. “A Quickstart Guide to Positioning.” 15 March 2021. Accessed 11 September 2026.
• Romy. “What's an ICP, and Why Do You Need One?” 26 August 2026. Reviewed 28 August 2026.
• Romy. “Who Actually Buys This? A 20-Minute Way to Name Your Buyer.” 19 June 2026. Updated 20 June 2026.
• Romy. “How Granular Should Your ICP Be When You're the Only One Selling To It?” 23 July 2026. Updated 27 July 2026.
• Strategyzer. “The Customer Profile.” Accessed 11 September 2026.
• Strategyzer. “5 Common Mistakes to Avoid When Using the Value Proposition Canvas.” 19 February 2015. Accessed 11 September 2026.
• Strategyzer. “How to Capture Customer Jobs, Pains, & Gains That Aren't Subjective.” 25 July 2016. Accessed 11 September 2026.
• Stripe Atlas. “Your First 10 Customers.” Accessed 11 September 2026.
• Stripe Atlas. “The SaaS Business Model.” Accessed 11 September 2026.