What is an ideal customer profile (ICP) in B2B?
Updated By the SalesOne research team8 min read
The short answer
An ideal customer profile (ICP) describes the companies most likely to buy from you, succeed and stay. In B2B it is written at the company level: industry, size, location, ownership and other checkable facts, plus the events that make a purchase likely now, the roles that decide, and the companies you exclude.
What is an ideal customer profile?
An ideal customer profile is a description of the type of company that gets the most value from what you sell and gives you the most value back: it buys, it succeeds, it renews and it refers others. It is not a list of companies. It is the set of rules that decides which companies belong on a list, and the first input to prospecting.
Most ICPs stop at firmographics, such as “US manufacturers with 200 to 2,000 employees”. That is a useful start, but it describes thousands of companies and says nothing about which of them to contact this month. A working ICP adds timing (buying signals with windows), people (the buying committee) and boundaries (exclusions).
- Key takeaway: an ICP describes companies; a persona describes people.
- Key takeaway: write every criterion so someone else could check it on public sources.
- Key takeaway: add signals and windows, or the ICP cannot tell you when to reach out.
What does ICP mean in sales?
In sales, ICP means ideal customer profile: a written description of the companies most likely to buy from you, succeed and stay, used to decide which companies to pursue.
Sales teams use the term as a filter. When a seller says an account is “in ICP”, they mean the company meets the profile’s rules, such as industry, size and location; “out of ICP” means it fails at least one. Professional services firms often call the same thing an ideal client profile.
- ICP vs persona in one line: the ICP picks the company; the persona describes the person you talk to there.
- An ICP in one sentence: US manufacturers with 200 to 2,000 employees, independent or PE-backed, with a new CFO in the last 180 days.
- Where it is used: building a target account list, deciding how to qualify leads, and ranking companies with account scoring.
What is the difference between an ICP and a buyer persona?
An ICP describes the companies to pursue; a buyer persona describes the people inside them. You need both, in that order, because a perfect persona at the wrong company is still the wrong account.
The two are built from different evidence. An ICP comes from company facts and your win and loss data: which kinds of companies bought, succeeded and stayed. Personas come from interviews, call notes and role research: what a head of procurement or a CFO cares about and how they buy. In B2B the ICP narrows the market first, and personas then shape who you contact at each account and what you say to them.
| Item | Ideal customer profile | Buyer persona |
|---|---|---|
| Describes | A type of company | A type of person |
| Built from | Company facts, signals, win and loss data | Interviews, call notes, role research |
| Typical contents | Industry, size, location, ownership, signals, exclusions | Title, goals, pains, objections, channels |
| Used for | Choosing and ranking accounts | Writing messages and choosing who to contact |
| Question answered | Which companies should we pursue now? | What does this person care about? |
What should a B2B ICP include?
A complete ICP has five parts: company criteria, buying signals, the buying committee, exclusions and target groups. The first is the one most teams already have; the other four turn it from a description into something research can apply.
| Part | What it contains | Example |
|---|---|---|
| Company criteria | Must, prefer and exclude rules on checkable facts | Must: US headquarters. Prefer: PE-backed. Exclude: in bankruptcy |
| Buying signals | Event types that predict a purchase, each with a time window | New CFO, 0–180 days; acquisition, 0–730 days |
| Buying committee | The roles to find and the minimum to count an account as reachable | Decision maker and champion required |
| Exclusions | Companies never to research as prospects | Existing customers, open deals, competitors, partners |
| Target groups | Segments that share most rules but differ in some | Mid-market carriers; enterprise shippers |
Company criteria: must, prefer, exclude
Mark every criterion with how strictly it applies: a must rule rejects any company that fails it, a preference only raises the rank, and an exclude rule rejects any company that matches it. Also decide what happens when a fact cannot be found. Private companies rarely publish revenue, so a revenue rule needs a substitute, such as employee count, or a rule to keep and flag.
Buying signals and the committee
Choose the event types that came before a purchase in your own wins, such as a new leader, an acquisition or a compliance deadline, and give each a window in days. SalesOne’s public library lists 43 signal types in 10 groups with typical windows. Then name the committee roles, such as decision maker, economic buyer and champion, and the minimum needed to count an account as reachable.
How do you build an ideal customer profile?
Build it from the customers you have, not the market you want. Start with your best customers and your losses, then turn what they share into rules. If what you sell and the problems it solves are not written down yet, start from your company profile.
- List your best customers: the ones that bought quickly, got results, renewed and expanded. Ten to twenty is enough to start.
- List lost deals and churned customers with the reason. They show what to exclude.
- Find the facts the best customers share and that you can check from outside: industry, size, location, ownership, business model.
- Look at what happened at each best customer in the months before they bought. Those events are your candidate signals.
- Write the rules: must, prefer and exclude, each with what to do when the fact is unknown.
- Set a window for each signal, based on how long after the event your past deals started.
- Name the committee roles and the minimum needed.
- Mark each rule as backed by data, stated by the team, or an assumption to test, and review the assumptions after the first research run.
The ICP is the first of six decisions in a B2B sales strategy; the others are timing, people, message, channel and measurement.
How do you identify your ICP when you are just starting?
When you are just starting, identify your ICP from the problem you solve rather than from customers you do not have yet: decide which companies feel that problem most, what event makes it urgent, and who owns it, then test that guess on real companies.
- Write down the problem in one sentence, in the buyer’s words, and the job title of the person who lives with it.
- Name the kind of company where that problem costs the most: its industry, size band and business model. Pick one segment narrow enough that you could list 100 companies in it.
- Name the event that makes the problem urgent, such as a new leader, a new site, an acquisition or a system change, and give it a window. This is what tells you when to reach out.
- Talk to 10 to 20 people in that role before you build anything bigger. Ask how they handle the problem today and what happened just before they last bought something for it.
- Use the customers and near-misses you already have, even two or three. Note what they share, and what the deals you lost had in common.
- Research a first batch of companies against the rules and read the rejects. The rules that turn down most companies are the ones to question first.
Expect most companies to fail. In SalesOne’s own runs, S1 agents checked 1,051 companies against sales profiles and kept 249, and every one of the 802 rejects kept its reason (how we measure). A first ICP that keeps almost everything is too loose to guide anyone. Founders still selling will find the next steps in founder-led sales, and the ICP template gives the rules a shape you can test.
What are examples of an ideal customer profile?
Two illustrative B2B examples follow: one for a company selling procurement software to mid-sized manufacturers, and one for a company selling security compliance software to growing software companies. Both use the same five-part structure, which is what makes them comparable and easy to test.
| Section | Procurement software for manufacturers | Compliance software for software companies |
|---|---|---|
| Must | US headquarters; manufacturing (NAICS 31–33); 200–2,000 employees | US or Canada headquarters; software company selling to businesses; 50–500 employees |
| Prefer | PE-backed or founder-led; three or more sites | Venture-backed, Series A to C; sells to enterprise customers |
| Exclude | In bankruptcy; subsidiary of a group that buys centrally | Layoffs announced in the last 90 days |
| Signals | New CFO or head of procurement (0–180 days); acquisition to integrate (0–730 days); job posts naming a procurement transformation (0–90 days); margin pressure in results (0–180 days) | Funding round (0–180 days); new CTO or head of security (0–90 days); job posts naming a SOC 2 audit (0–90 days) |
| Committee | Decision maker (CFO or COO) and champion (head of procurement) required; evaluator if the role exists | Decision maker (CTO) and champion (head of security) required; economic buyer (CFO) if the role exists |
| Exclusions | Existing customers, open deals, competitors, partners, do-not-contact list | Existing customers, open deals, competitors, partners, do-not-contact list |
How do you know your ICP is working?
Run research against it and read the rejects, then compare accepted accounts with meetings held. The rejects show which rules do the work; the meetings show whether those rules predict outcomes. In SalesOne production runs, about three of every four companies fail the profile they were checked against (what 802 rejects show).
If most companies fail one must rule, check whether that rule reflects how you really win. If most good-fit companies have no signal inside its window, the windows may be too short or the signal list too narrow. Accepted accounts that never convert point to a rule that is missing.
- Acceptance rate: the share of researched companies that qualify. Very high usually means the rules are too loose.
- Reject mix: which rules reject the most companies, and whether that matches your expectations.
- Meetings per accepted account, by segment and by signal type.
What are common ICP mistakes?
Most ICP mistakes make the profile either too vague to apply or too static to stay true. Each one below leads to research that cannot be checked or outreach to companies that were never going to buy.
- Writing the ICP from the market you want rather than the customers you have.
- Using criteria nobody can check, such as “innovative” or “growth-minded”.
- Leaving out timing, so every good-fit company looks equally ready.
- Forgetting exclusions, and prospecting your own customers or partners.
- Never revisiting it. An ICP is a set of hypotheses; research results should change it, through regular ICP refinement.
How does SalesOne use an ICP?
Most tools turn an ICP into database filters: industry, size and location. SalesOne starts one step earlier. S1 agents first learn your own company from your material, what you sell, who already buys and the proof behind it, and draft the profile from that. Your team edits and confirms it, and every accept and reject is then judged against it with a written reason.
In SalesOne, you build the ICP from your own company in the ICP builder, and it is structured: company criteria marked must, prefer or exclude; buying signals chosen from 43 types in 10 groups, each with a window; a buying committee of six roles; exclusions; and target groups that override parts of the profile.
S1 deep research agents work from it directly, so the same rules decide every accept and every reject. Your ratings and reject reasons then show which rules to change. S1 agents do the homework. Your team builds the relationship.