
An ICP, or ideal customer profile, describes the type of company most likely to buy from you, get value and stay. It is about accounts, not people. A good one is short enough that an SDR can apply it in seconds and specific enough that it removes most of the market from the call list.
These terms get mixed up, and the confusion leads to messy lists.
You need the ICP first. A perfect persona at the wrong company is still a wasted call.
Persona work is where many teams stop, and it is why their lists fail. Marketing ends up with five personas, while reps have no rule for which accounts to open first. Fix the account filter, then the persona.
Think in three layers.
A fourth piece, the persona, comes after. It decides who you ask for once you have the account.
Do not start in a workshop. Start with your last 20 to 30 closed-won deals, and your closed-lost ones.
If you have fewer than 20 wins, do not pretend the data is bigger than it is. Combine wins with your best pilots, lost deals and the conversations your reps found easiest, and label the result as a hypothesis. A rough ICP tested on 200 calls beats a polished one nobody checked.
Here is an illustrative example, not a real customer: a French B2B software company with 20 to 200 employees, an outbound team of 3 to 15 reps, a CRM with call activity logged, and a recent hire or funding event. Disqualify companies with no sales team and agencies that resell your category.
An ICP is a hypothesis until real calls test it. Your dialing data can do that in a few weeks.
Tag every call by segment and persona, then compare:
If one segment gives you twice the meetings per 100 dials, move it up. If a segment connects well but never books, revisit the pain you assumed. A persona that picks up is not the same as an account that fits.
Watch for one trap: an ICP that fits your existing customers too tightly. If every win came from one network of referrals, the profile will reflect your network, not your market. Test one adjacent segment each quarter with a small block of calls, and let the data say whether to widen the profile.
An ICP that lives in a slide deck changes nothing. Put it where the work happens.
We treat the ICP as a filter on the signals in your CRM. A website visit from an account outside your ICP is noise, while the same visit from a Tier 1 account is a call for today. The signals do the timing, and the ICP does the filtering.
The dialer then puts those accounts in a queue, with parallel dialing to give reps about 5 times more conversations. Because every call is logged by segment and persona, you can check the ICP against what really happens on the phone.
Your ICP is only right if calls prove it. This week, pull your last 100 dials, tag each by segment and persona, and count conversations and meetings for each. Pick the best segment, write its ICP in five lines and call it first on Monday.