Ideal customer profile (ICP)

What an ideal customer profile contains

An ICP describes companies, not people. A usable one fits on a page and has three parts.

1. Firmographic traits. Industry, headcount, revenue band, geography, ownership (founder-led, PE-backed, public).
2. Situation traits. The conditions that create the need: a system they have outgrown, a regulation they now fall under, a team that just doubled.
3. Exclusions. The companies that look right on paper but churn, stall in procurement or never expand. Writing these down saves more time than the inclusions do.

The profile comes from evidence. Start with the customers who pay the most, renew, and grow, then find what they have in common that weaker customers lack. It is a form of market segmentation narrowed to one segment: the one worth most of your effort. McKinsey's checklist for B2B growth asks the same of sales teams in money terms: quantify the spend opportunity of every customer and prospect, then prioritize. Once written, the ICP becomes the reference for every fit score and for account scoring across a target list.

How to build an ICP from customer data

1. Pick the outcome. Decide what "best" means: net revenue retention, gross margin, time to close, or a blend. Choose before you look at the list, or you will pick the measure that flatters your favourite accounts.
2. Split the customer base. Rank customers on that outcome and compare the top quarter to the bottom quarter.
3. Find the traits that separate them. Look for traits common in the top group and rare in the bottom. A trait shared by everyone tells you nothing.
4. Write it down with exclusions. State the profile in plain sentences that a new sales hire could apply without asking.
5. Test it on lost deals. If many deals you lost also match the ICP, it is too broad.
6. Review it on a schedule. Re-run the split every two or three quarters. Products and markets move, and an ICP written for the first 50 customers rarely fits the next 500.

Worked example: Ledgerly's ICP

Ledgerly is a fictional Series A fintech that sells reconciliation software. It has 140 customers. The team ranks them by net revenue retention and compares the top 35 with the bottom 35.

Geography does not separate the groups, so it stays out. The resulting ICP: "Companies with a finance team of 5 to 20 that run two or more payment processors. Exclude companies in the middle of an ERP migration." That last line is the useful surprise. Those buyers signed quickly and then had no capacity to roll the product out.

ICP vs buyer persona

Teams need both, in that order. The ICP narrows the market to accounts; personas shape the conversation once you are inside one. In account-based marketing, the ICP decides the account list and personas decide who gets which message. Personas also feed lead scoring, where job title and seniority earn points. When the account list goes to outbound, cold outreach scoring ranks the contacts nobody on the team knows yet.

Why investors ask to see a company's ICP

In commercial diligence, a venture or private equity team often asks a company to describe its ideal customer and then checks the answer against the customer list. A sharp ICP that matches who actually renews is evidence of a repeatable motion. A vague one ("mid-market companies that care about efficiency") usually means the company sells to whoever answers. That gap shows up later as weak retention or long sales cycles, and it is part of how investors judge go-to-market quality.

Investors run the same logic on themselves. A fund's thesis is its ICP for deals: the companies it is built to back. Thesis fit scores a company against it, and market mapping lays out every company in a space so the fund can see which ones match. The discipline is the same: write the profile from evidence, keep exclusions, and test it against the ones you lost. McKinsey's 2026 study of B2B growth leaders describes AI agents that enrich account profiles with firmographics and score each opportunity on fit, timing and likelihood to convert. A written ICP is what the fit part gets scored against.

Common ICP mistakes

• Writing it from logos you want. An ICP built on aspiration describes the customers you wish you had. Build it from the ones who stayed and grew.
• Only firmographics. Industry and headcount are easy to filter but rarely explain why a customer buys. Add the situation that creates the need.
• No exclusions. Without them the ICP describes half the market and ranks nothing.
• One ICP for every product line. Different products often have different best customers.
• Never revisiting it. The profile that fit your first customers may not fit the next wave.

How ScoringFactory approaches it

ScoringFactory does not sell sales software. It applies the ICP idea to investing: it learns a fund's or firm's profile from the companies it backed and passed on, scores each new company against that bar, and cites every score to the record behind it. The team still makes the call. If you are mapping a market against a mandate, talk to the founders.