The target
Cold email works because you research each account before you write. That only holds if your ideal customer profile is countable: a list you could research, not a market you can only blast. Here is how to define one that actually replies.
The ICP is not who you would love to sell to. It is the shape of the accounts that already bought and stayed: their size, their structure, the trigger that made them ready. List the last ten good customers and find what they share.
If your ICP has 50,000 accounts, it is not an ICP, it is a demographic. Cold outbound works because you research each account before you write, and you cannot research 50,000. A real ICP is a list you could put in a spreadsheet: often a few hundred to a few thousand accounts.
“B2B SaaS, 50 to 200 staff” describes a category, not a moment. The moment is what makes cold email land: they just raised, they are hiring SDRs, they switched a tool, they opened a market. The firmographic says who; the trigger says now.
For each ICP, know the person who signs and the function that gates. In much of B2B, ownership decides suppliers and a local executive cannot; procurement is a separate gate with its own address. If your email reaches the wrong altitude, the best case is a forward, so write for the forward.
Run more than one ICP as separate campaigns, not one blended list. Measure each on replies, meetings and closes, never opens. The ICP that answers gets the volume; the one that goes quiet gets cut with the reason written down. Your ICP is a hypothesis until the replies confirm it.
The constraint nobody mentions
When we scope a programme, the binding number is not the budget or the copy. It is how many accounts genuinely fit the ICP. That total addressable list is usually smaller than people expect, and it sets the ceiling on everything downstream. It is also why we do not A/B test copy on small markets: at cold email’s reply rates, splitting a few hundred accounts several ways teaches nothing. Below real volume, you write one good sequence and put the effort into research per account, which is exactly where a tight ICP pays off. This is how we pick targets when we run outbound as an agency.
Questions
For cold outbound, an ICP is a countable list you can research account by account, using a firmographic filter, a trigger, and a named buyer, not a demographic.
Start from the customers who already bought and stayed, not the ones you wish you had. Find what they share: size, structure, and the trigger that made them ready. Turn that into a countable list of accounts, name who signs and who gates in each, then run it as its own campaign and let replies, meetings and closes confirm or kill the hypothesis.
Narrow enough to research. The binding constraint on a cold outbound programme is usually how many accounts genuinely fit: often a few hundred to a few thousand per ICP, not tens of thousands. A tight ICP you can research per account beats a broad one you can only blast, because relevance and checkable facts move reply rates far more than volume does.
Often yes, but run them as separate campaigns, never one blended list, so you can tell which is actually working. Measure each on replies and meetings, put volume behind the one that answers, and cut the one that does not. An ICP is a hypothesis; the data decides.
The ICP is the company you target; the persona is the person inside it you write to. You need both: the ICP tells you which accounts make the list, and the persona, including who signs and who gates, tells you how to write and to what altitude. Reaching the right company at the wrong person usually means, at best, a forward.
Targeting is where the cold email playbook starts. Once the ICP is tight, the subject line and the ask do the rest.
EMPRA