The ideal customer, and the customer to avoid

An ideal customer profile describes the companies most likely to buy, succeed with the product and stay. Everything downstream depends on it: which data you buy, which signals you watch, what you write. A vague ICP makes every later step expensive and every message generic.

Why the ICP comes first#

It is tempting to start by pulling a big list. But the list is only as good as the definition behind it. A tight ICP cuts data costs, because you enrich fewer accounts. It raises reply rates, because the message can be specific. And it makes sales happier, because the meetings they get are with companies that can actually buy.

Start from the deals you won#

Opinions about the ideal customer are cheap. The CRM holds evidence. Pull the last 50 to 100 closed-won deals and, if possible, the ones that renewed or expanded. Look for what they have in common. Then do the same for closed-lost deals and customers who churned. The contrast is usually more informative than either list alone.

1. 1

   Export won, lost and churned deals

   With company size, industry, region, deal size, sales cycle length and the reason won or lost, if recorded.
2. 2

   Enrich what's missing

   Headcount, funding stage, tools they use, growth rate. Fill gaps before drawing conclusions.
3. 3

   Look for over-representation

   If 40% of wins are logistics firms but logistics is 5% of the market, that's a signal.
4. 4

   Check value, not just volume

   Segments with fewer wins but bigger deals or better retention may matter more.
5. 5

   Talk to sales and customer success

   Ask who was easy to sell to, who got value fast, and who was painful. Their stories explain the numbers.

Three kinds of attribute#

Firmographics tell you who could buy. Situations tell you who is likely to buy soon. Chapter 5 is about turning situations into signals you can monitor.

Write the anti-ICP too#

Equally useful is a short list of companies you will not target, even if they look like a fit on paper: segments with long sales cycles and small deals, companies too small to pay, competitors, existing customers, regions you can't serve, industries where the product has failed before. Excluding them early saves data costs and protects reps' time. It also stops you sending embarrassing messages to existing customers.

Who you leave out does as much for reply rates as who you put in.

Make it filterable#

An ICP that a database can't filter on is a slogan. "Innovative growing companies" can't be searched. "Software companies in the UK and Ireland, 50 to 250 employees, headcount up more than 15% in a year, selling to enterprises" can. For each attribute, write down the exact field and values you will use in your data sources, and note where an attribute can only be checked by research or AI (more on that in chapter 10).

• Use ranges, not adjectives: 50–250 employees, not 'mid-size'.
• Name the source for each attribute, because different databases define industry differently.
• Mark which attributes are hard filters (must have) and which are scoring factors (nice to have).
• Version the ICP. When you change it, note why and what the data showed.

From company to person#

Inside each company several people are involved in a purchase. Write a short persona for each role that matters: what they are responsible for, what they are measured on, what they worry about, and how the product helps. Usually there is a person who feels the problem, a person who owns the budget, and someone who can block the purchase, such as IT or legal. Your first message usually goes to the person who feels the problem most.