Ideal customer profile (ICP) for B2B: template and examples 2026
2026-10-15 · updated 2026-10-15 · 11 min read · Adam Axelsson, founder of Revexa
An ideal customer profile (ICP) describes the type of company that gets the most value from what you sell and is worth the most to you, so it decides which companies you pursue and which you skip. A buyer persona describes a person inside those companies: their role, goals and objections. The fields that matter most in a B2B ICP are business model, size, market, buyer roles, the triggers that make the problem urgent, and who to exclude.
Checked on 15 October 2026 against each vendor's and publisher's own pages. Revexa builds Leo, one of the tools below. Every company example on this page is fictional, and numbers marked "our rule of thumb" are our estimates, not benchmarks.
ICP vs buyer persona: what is the difference?
| Ideal customer profile (ICP) | Buyer persona | |
|---|---|---|
| Describes | A type of company | A role inside that company |
| Answers | Which companies do we pursue, and which not? | Who do we talk to, and what do we say? |
| Typical fields | Business model, industry, size, market, technology, triggers, exclusions | Title, seniority, goals, KPIs, problems, objections |
| Without it | You reach the right people at companies that cannot buy | You reach the right companies with a message that fits nobody |
Outbound needs a third piece: a signal that says when. The ICP says who fits, the persona says who to talk to, and the signal says why now.
How to define an ICP for B2B SaaS, step by step
Step 1: If you have customers, start with won and lost deals
Export won and lost deals from the last 12 to 24 months (our suggested window) with the reasons your CRM stores: HubSpot has default "Closed won reason" and "Closed lost reason" properties (HubSpot), and Pipedrive asks for a lost reason when a deal is lost (Pipedrive). Mark the 10 to 20 customers you would clone (our rule of thumb), then answer:
- What do the best customers share that the rest do not? Look at business model, size, market and team, not only industry.
- What happened just before they bought: a new leader, a round of hires, funding? Those events become your triggers.
- Who found you, who owned the problem and who signed?
- Why did lost deals stop: budget, timing, a competitor or no real problem?
- Which customers churned or cost more to serve than they paid? What they share is your negative ICP.
The same export is where segmenting your CRM list starts.
Step 2: If you are early, start from your website and offer
With fewer than about ten customers, patterns are noise (our rule of thumb), so write a hypothesis. Your website answers half of it: who the copy speaks to, which integrations and logos you show, what your pricing assumes. Then answer:
- Which problem do you solve, and for whom is it expensive enough to pay to fix?
- What must be true inside a company for the problem to exist: a business model, a team, a process, a tool?
- Who feels the pain every week, who controls the budget, and what do they use instead of you today?
- Which companies would you turn down even if they wanted to buy?
Step 3: Test the ICP against reality
- Hand-check a sample. Pull 50 companies that match on paper and check 20. If fewer than half have the problem, the profile is too broad (our rule of thumb).
- Run segments side by side for four to six weeks (our rule of thumb) and compare replies and meetings per segment.
- Ask in every first call what made them look now. The answers sharpen your triggers.
- Study the misfits: customers won outside the profile show where it is too narrow, churned ones inside it where it is too loose.
ICP template for B2B (with a fictional example)
Copy the table and replace the last column, which describes a fictional B2B SaaS company selling a customer onboarding platform.
| Field | What to write | How to check a company | Fictional example: customer onboarding platform |
|---|---|---|---|
| Industry | The industries you serve, in words rather than codes | Website, LinkedIn industry | B2B software sold to other companies |
| Business model | How the company sells and earns money; this field most often decides fit | Pricing page, demo or free-trial button, customer logos | Sales-led, annual contracts, a hands-on onboarding project for every new customer |
| Size | A headcount range and the team that must exist | LinkedIn headcount, annual reports | 50 to 500 employees, a customer success team of at least three |
| Market | Countries you can sell to, serve and legally contact | Headquarters, offices, website languages | UK, Nordics and Benelux where each country's email rules allow; Germany only via inbound and partners, as advertising email there needs prior express consent (UWG §7) |
| Buyer roles | Problem owner, signer, users, any blocker | LinkedIn titles, job ads | Owner: Head of Customer Success. Signer: COO, or the CEO below 200 employees. Users: onboarding managers. Blocker: IT security |
| Technology | Tools you integrate with or replace | Integration pages, job ads, technology lookups | HubSpot or Salesforce, a helpdesk such as Zendesk, onboarding in spreadsheets today |
| Triggers and buying signals | Events that make the problem urgent now | Job boards, funding news, LinkedIn job changes | Hiring onboarding managers, a new VP of Customer Success, a Series A or B, a new enterprise plan |
| Exclusions | Companies and people you never contact | Your CRM and suppression lists | Customers, open deals, competitors, partners, everyone who opted out |
| Negative ICP | Companies that look like a fit but stall, churn or cost too much | Churned customers, lost deals | Self-serve-only SaaS with no onboarding, agencies selling onboarding as a service, companies under 30 employees |
The same profile in one sentence: sales-led B2B SaaS companies with 50 to 500 employees in the UK, the Nordics and Benelux that run a hands-on onboarding for every new customer, where the Head of Customer Success owns the problem; top priority when a trigger has fired in the last 90 days.
Mark each field as a must-have or a plus (here: business model, size and market are must-haves) and keep exclusions apart from the negative ICP. Exclusions are hard rules, some of them legal: under GDPR article 21(3), after an objection to direct marketing the data "shall no longer be processed for such purposes" (GDPR). The negative ICP is a judgment you revisit. This is not legal advice.
How do I find companies that match my ICP?
Turn each field into a filter, in this order:
- Filter on what databases know: size, headquarters, industry and headcount growth are standard filters in Apollo, LinkedIn Sales Navigator and Clay.
- Check the business model on the website. Clay's documentation says its industry categories "map to the categories companies self-select on LinkedIn" and recommends description keywords, which "let you narrow by what a company actually does" (Clay), and Apollo's AI research can tell "whether a company is B2B or B2C" (Apollo).
- Add a trigger. Sales Navigator filters accounts on headcount growth over 12 months and open job opportunities, and leads on job changes in the last 90 days (LinkedIn). Apollo filters on job postings, funding and headcount growth.
- Remove exclusions first: customers, open deals and opt-outs. Clay, for one, can exclude "up to 150,000 companies or people" from a search.
- Test coverage in your market. Clay expects 15 to 25 percent lower coverage than in the US for EMEA and APAC. Run any database against 20 companies you know before paying for a year.
Fit and timing: using your ICP with buying signals
The ICP tells you who fits, not when. The 95:5 rule from LinkedIn's B2B Institute and Professor John Dawes of the Ehrenberg-Bass Institute (2021) says "95% of your potential buyers aren't ready to buy today", based on how rarely companies buy: "80% of companies change banking services once every 5 years" (B2B Institute). It is a rule of thumb, not a measurement, but even a perfect-fit list is mostly out of market today.
Signals show which of them are moving. In Bloomberry's analysis of 1 million software purchases from March to September 2025 (its own purchase data, mostly companies with 200 to 1,000 employees), companies that hired a VP made 28% more software purchases, headcount growth of 20% or more went with 38% more and a recent funding round with 25% more among startups over six months, while a 20% rise in job openings went with only 7% more (Bloomberry). These are correlations, not proof.
In outbound, put fit first: only companies that pass the must-have fields go on the watch list, since a signal outside the ICP is noise. When a trigger fires, write to the role that owns the problem and open on the problem, not the signal.
Guides to the signals: 30 B2B buying signals, companies hiring SDRs, recently funded startups, signs a company is switching software and intent data. For a weekly routine on a one-page profile, see founder-led outbound.
Common ICP mistakes
- Too broad. "B2B companies with 10 to 1,000 employees" filters nothing and produces irrelevant outreach. In a Gartner survey of 632 B2B buyers (fielded August to September 2024, published June 2025), "73% of B2B buyers actively avoid suppliers who send irrelevant outreach" (Gartner). If your profile does not rule out most of your market, it is a market, not a profile.
- Only firmographics. Industry, size and location say what a company is, not whether it has your problem or how it buys; no industry label separates sales-led from self-serve SaaS. Add business model and triggers.
- No exclusions. Without them you write to customers, open deals and people who said no. Some exclusions are legal: GDPR article 21(3) for objections, and UWG §7 in Germany for advertising email without consent. This is not legal advice.
- Never updated. A profile written before your last pricing change or launch describes a different customer. HubSpot recommends reviewing the ICP "at minimum once per quarter" (HubSpot).
Tools that help you build and use an ICP
Listed alphabetically. Prices are from each vendor's own page on 15 October 2026, excluding VAT.
| Tool | What it does for your ICP | Price from | Best for | Wrong choice if |
|---|---|---|---|---|
| Apollo | Company search on employees, industry, technologies, funding, headcount growth and job postings; lookalikes; AI research on fit. Some filters need a paid plan | Free plan; Basic $49 a seat a month, billed annually | A first filtered list on a small budget | Fit depends on business model and nobody has time to check websites |
| Clay | Description keywords and AI filters for business types, exclusion lists, and Claygent, "Clay's AI agent with access to web research" | Free plan; Launch $167 a month billed annually, $185 monthly | Teams with someone who builds and maintains workflows | Nobody owns the build, or your accounts are mostly outside the US |
| CRM reports (HubSpot, Pipedrive) | Won and lost deals by company field, lost reasons, and in HubSpot an "Ideal Customer Profile Tier" property | HubSpot free CRM; Pipedrive from €14 a seat a month, billed annually | Building the ICP from your own deals | You have closed too few deals for patterns, roughly under 20 (our rule of thumb) |
| Leo by Revexa (ours), a buying-signal agent | Proposes the ICP from your website on day one: industries, size, markets, buyer roles and companies that do not fit; you add your own words in free text, with no questionnaire. Judges each company's business model before anything is sent, then watches buying signals and drafts LinkedIn and email messages for your approval | €89 a month in Europe, $99 in the US and rest of world, 999 SEK in Sweden; 7 days free | Small B2B teams that want the ICP, signals and drafts in one place | You want a database to export lists or website visitor identification, you sell by phone, SMS or WhatsApp, or your market is mainly Germany (no cold messages there, UWG §7) |
| LinkedIn Sales Navigator | Account filters for headcount, 12-month growth, industry, headquarters and job opportunities; lead filters for title, seniority and recent job changes | Core US$119.99 a month or US$1,079.88 a year (LinkedIn calls its prices estimates) | Finding the buyer roles at accounts that fit | You need to filter on business model, such as B2B versus B2C, which is not among the filters LinkedIn lists |
Common questions
How do I define an ideal customer profile for B2B SaaS?
With customers, find what your best won deals share and what lost and churned deals share; without them, write a hypothesis from your website and offer. Then test it on a hand-checked sample.
What should an ICP include?
Industry, business model, size, market, buyer roles, technology, triggers and buying signals, exclusions and a negative ICP. Business model, size and market are usually the must-haves.
What is the difference between an ICP and a buyer persona?
An ICP describes a type of company; a buyer persona describes a role inside it. The ICP chooses the companies, the persona the person and the message.
How often should you update your ICP?
Review it every quarter against won, lost and churned deals, as HubSpot recommends, and rewrite it after a pricing change, new product or new market.
How do I find companies that match my ICP?
Filter a database on size, market and industry, check the business model on each website, add a trigger such as hiring, remove exclusions and hand-check a sample before sending.