ICP (ideal customer profile)

An ICP describes a typical account. It lists what you can see from the outside: industry, team size, country, business model, growth stage. In B2B it almost always comes with a role too: the person inside that company who has the problem you solve and the authority to fix it.

Don't mix it up with a buyer persona. A persona tells the story of a person, their day, their objections, what they read. An ICP is a filter. Quick test: if you can't tell within ten seconds of reading a LinkedIn profile whether someone fits, your ICP is written like a persona.

In outbound prospecting that filter matters more than anywhere else. Every invitation and every message draws on a quota LinkedIn limits. A vague ICP fills that quota with people who will never buy, and the campaign numbers slide without anyone understanding why.

Start with your best current customers, the ones who signed quickly, pay without renegotiating and stay. List what they share that is visible from the outside. Two or three traits usually come back: a team size, a type of clientele, the role that carried the decision.

Then write the ICP as a sentence a stranger could apply, for example "founders of marketing agencies with 5 to 30 people in the UK that sell retainers". Every word should be checkable on a profile or on the company website.

Keep hard criteria apart from bonus signals. A need, a buying intent or a budget can't be read on a profile. Turn them into conditions and you'll reject your best buyers for lack of proof. Keep them as signals that move someone up the queue, never as a filter.

Finally, note what disqualifies someone for sure: competitors, students, inactive profiles. Keep that list short and factual, one line per type of profile, or it ends up excluding people you actually wanted to reach.

Judge an ICP on the numbers of each segment: invitation acceptance rate, reply rate, meetings booked, then customers signed. Compare a segment with itself or with its industry, not with a global average. In the MimikFlow 2026 Observatory, 30.2% of 50,930 invitations at least 30 days old were accepted, but the rate ranged from 20.4% in real estate to 36.4% in data and AI.

The role you target shifts the benchmark as well. In the same study, consultants and freelancers replied 57.1% of the time, while directors and department heads accepted only 24.5% of invitations. A low rate on a director audience doesn't automatically mean a bad ICP.

Don't conclude from a handful of profiles. Wait for a few hundred invitations per segment before deciding, and look at meetings and signed deals first. A segment that replies a lot but never buys is a chatty segment, not a good ICP.

The most common one is an ICP that is too broad, like "SMBs" or "entrepreneurs". It produces generic messages, because you can't say anything specific to someone you know nothing about.

The second is writing invisible criteria as conditions: "wants to grow", "has budget", "is looking for a solution". Nobody writes that on their profile. Those words belong on the bonus signals line.

Third trap: treating the ICP as a list of job titles. The same role is called Head of Growth, Acquisition Lead or Marketing Director depending on the company. Describe the function and the situation, not just the title.

The last mistake is freezing it. Your offer changes, so do your customers. Reread your ICP when one segment clearly stands out in your meetings or when your offer changes.

In MimikFlow you describe your target in a few lines, in your own words. The platform turns that description into searches, reads every profile it finds and gives it a relevance rating along with the reason for that rating. Profiles that don't fit are not contacted.

The same description then guides the first message, the follow-ups and the AI Setter's replies. One useful caveat: in the 2026 Observatory, the relevance rating did not predict who would reply, with about two points between a rating of 6 and a rating of 9. The ICP tells you who to talk to. The message and the timing are what get a reply.