ideal customer profile (ICP)
Also called: ICP, ideal customer profile
A description of the kind of company that gets the most value from the product and is worth the most to the seller: industry, size, situation and signals that can be checked. It is built from real won and lost deals, not from wishes.
An ideal customer profile (ICP) describes the kind of company that gets the most value from a product and is worth the most to the seller: where deals are won, where clients stay, and where serving them pays. It is a description of companies, not of people; the people inside them are mapped separately (the buying committee, buyer personas).
A useful ICP is built from the seller's own history: won and lost deals, who renewed and who left, how long deals took and what each client costs to serve. Typical lines cover industry, size, the reason the company needs the product, deal size, speed of buying and cost to serve. Every line should be checkable before the first call, for example "IT services firms of 60 to 300 people where 15 to 60 staff talk to foreign clients in English", not "innovative companies that value learning".
An ICP also says whom not to chase, through disqualifiers drawn from lost or unprofitable deals. When the data is thin (a dozen wins is common early on), mark low-confidence lines as such and review the profile every quarter: it is a hypothesis that new wins and losses keep testing. Companies outside the profile can still be pursued, but as named bets with the condition that would move them in, not by default.
Example
Klasna for Teams talked to 40 companies in 2024–2026 and won 12 (30%). Revenue alone points to retail: its three clients brought 249,100 of 484,500 UAH of monthly revenue, 51.41%. Profit after the cost of serving tells the opposite story:
IT services are 29.6% of revenue (143,400 / 484,500) but 56.98% of profit (53,770 / 94,370); retail is 7.66% of profit (7,230 / 94,370). Retail lessons carried coordinator hours and paid no-shows: 71,300 UAH of service cost ate 90.79% of its 78,530 UAH contribution. And all three retail clients bought English as a staff perk; none is still a client.
So the first ICP line is about IT services firms whose staff use English with foreign clients, not about the biggest names. Logistics (four deals, none won) is not written off as an industry: the reasons behind the losses become disqualifiers, and Veloria, with a real work need, is pursued as a named bet with a paid pilot.
Common mistakes
- Building it from market size or wishes. Start from the clients who stayed and pay off, not from the longest list of companies.
- Ranking segments by revenue. Revenue hides the cost to serve; compare profit per client or per unit.
- Vague lines. "Innovative, growth-minded companies" fits everyone and can't be checked before a call.
- Writing off an industry for a few losses. Screen for the reasons behind the losses, not the label.
- Treating it as permanent. With a dozen wins, confidence is low; mark it and review the profile every quarter.
Learn it in the course
- An ideal customer profile from your best customers · Build an ICP from won and lost deals: industry, size, why they need the product, deal size, speed and cost to serve, plus the disqualifiers that save you time.
- Scoring segments and accounts: profit, not revenue · Turn ICP criteria into a score: the direction and scale of each criterion, weights that add up to 100%, profit per lesson after service cost, and scoring bottom-up from real accounts.
- Case: forty companies, who is our ideal customer? · Read Klasna for Teams' won and lost deals since 2024, rank segments by profit after service cost, and write the ideal customer profile, disqualifiers and bets for 2027.