What is an ICP (Ideal Customer Profile)?

An ICP (Ideal Customer Profile) is a detailed description of the company and buyer persona most likely to purchase your product, derive value from it, and become a long-term customer. It defines the firmographic, technographic, and behavioural characteristics of your best-fit accounts — and acts as the filter for every outbound, paid, and inbound motion you run.

What does an ICP include?

A complete ICP has two layers: company-level and contact-level.

The company layer covers firmographics: industry, headcount or revenue, geography, business model, funding stage, tech stack, and growth signals. A real example looks like this — "Series A–C B2B SaaS companies, 20–200 employees, US-based, running Salesforce or HubSpot." Specific enough to pull a list from Apollo or LinkedIn Sales Nav.

The contact layer covers the buyer: title, seniority, decision-making authority, budget ownership, and the problem they're paid to fix. Not "someone in sales leadership" — "VP Sales at a company that just hired its third SDR and has no SDR manager yet." That's a person with a specific pain, a budget to solve it, and a reason to respond to you today.

ICP vs buyer persona: what's the difference?

An ICP describes the ideal company. A buyer persona describes the ideal contact within that company. In practice, the two are often combined into a single profile, but the distinction matters for prospecting: the ICP filters which accounts to target, the persona filters which person at that account to reach out to.

Why does the ICP matter so much?

Every GTM motion downstream depends on the ICP being right. Outbound uses it to filter which companies to target and which contacts to reach. Paid uses it to set targeting parameters. Content uses it to choose topics. Get the ICP wrong and you're spending money talking to people who don't have the problem you solve — and wondering why nothing converts.

A vague ICP produces generic outreach. Generic outreach produces low reply rates, long qualification calls, and deals that stall at procurement. A tight ICP does the opposite — your message lands because it names a problem the reader actually has.

How do you build an ICP?

Start with your best existing customers. Which accounts got the most value fastest? Which had the shortest sales cycles? Which expanded or renewed? What do they have in common — industry, size, tech stack, the role of the buyer?

No customers yet? Start with a hypothesis: who is most likely experiencing the pain you solve, has budget, and has a decision-maker with authority to buy? Write it in one sentence. That's your starting ICP. Test it with outbound. Refine it based on what replies and converts.

How does the ICP connect to AI agents?

In an AI-native GTM system, the ICP lives as a typed record in the Team Brain — shared across every agent simultaneously. The SDR agent scores and filters prospects against it. The AE references it when qualifying replies. The Marketing Manager pulls from it to align content topics. The Ad Manager reads it to set targeting. One record. Four agents. No briefing required.

When the ICP changes — because you've learned something new from deals won or lost — updating the Team Brain propagates that change to every agent on the next cycle. The whole team realigns without a meeting.

How specific should an ICP be?

Specific enough to generate a list. If your ICP can't be used to search a prospecting database and return a finite, relevant list of accounts, it's too vague. "Growing B2B companies" is not a usable ICP. "Post-Series A B2B SaaS companies, 50–500 employees, with a VP Sales but no dedicated SDR team" is.

Narrow ICPs feel risky. They produce better results. You can always expand once you've proven traction in a segment — but starting broad means every message is generic and every call is a longer qualification process.