An ICP gives outbound a rule for choosing accounts before you spend time on them. Build it around companies you are best at winning, serving, retaining, and expanding.1 Firmographic and technographic data are a starting filter. Used alone, they leave you hoping to find good customers.2 Use the profile to decide who you research, contact, and leave alone.
Build the profile in work order
Start with the rule, then build the list. Each pass narrows who gets attention and makes the next research task easier.
Define the account outcome
Start with the result you need from an account after the sale. An ICP should answer who will buy, who will buy smoothly, which accounts can be implemented and serviced, and which will renew and expand.3
Write each answer as a check you can apply to a company. Ask what has to be true for the account to succeed after purchase. Include the conditions that make the deal workable for the customer and your team. Move on when the profile describes success beyond the signed contract.
Study actual customers
Use customer history to find the pattern before you name the market. Your most successful deployments are likely to represent the closest starting point for the ICP.4
Compare the customers that have been best for the business with those that have been hardest to win, serve, or retain. Reverse engineer why each group behaves that way before choosing the attributes to copy.5 After you build a few profiles, look for a pattern across them.6
For each account, record what made the purchase possible, what made delivery work, and what created friction. Keep asking whether an attribute explains the outcome or merely describes the company. A field belongs in the profile when it helps you predict fit or rule out a poor fit.
Draft the account rule
Turn the pattern into fields that someone can check without reading a paragraph of interpretation. Use industry, size, business model, technical environment, and buying triggers as candidate fields.7
Keep the rule to a short list of firmographic and technographic criteria, add explicit disqualifiers, and score and tier it in the CRM.8 Write disqualifiers as clear stops. If an account fails one, the reason should be visible without a debate about whether the company feels promising.
Ask which fields change the decision and which only make the profile sound complete. Remove the fields that do not change account selection. The finished rule should help you decide who belongs in the list before you research individual contacts.
Separate account fit from person fit
Keep company selection and person selection in separate layers. An ICP describes the company, while a buyer persona describes the person inside it who evaluates and buys.9
Build the ICP before the personas because it determines which accounts are worth pursuing.10 Once an account passes the rule, map the people involved in its purchase and tailor the conversation to what each person needs to assess. A strong persona cannot rescue an account that fails the company level criteria.
Test the rule with outbound
Your first profile is a hypothesis, so put it in front of real prospects instead of polishing it in isolation. Form a hypothesis for what the ideal customer looks like, then research prospects against it.11
Reverse engineer the pattern from prospects who took meetings, had painful discovery calls, or eventually bought, then look for more people who resemble them.12 Use outbound sales data as the main signal for what is working, since it is a better signal than leads from investors and friends.13
Watch for mismatches. If accounts that fit the rule ignore the problem while accounts outside it engage, inspect the rule before blaming the message. Change the field, disqualifier, or segment that fails the test, then run the revised rule through outbound again.
Put the rule into account selection
An ICP changes execution only when it controls who gets time and research. Use it to decide which accounts deserve time and money spent trying to win them.14
Give each account a reason for its fit and a visible reason it could be excluded. Keep the score close to the account record so selection, research, and follow up use the same rule. When the list grows, tier it by the strength of the fit and work the strongest matches before weaker ones.
What not to do
These mistakes make the profile look precise while weakening account selection.
- Do not create a fictional idealized profile. Use customers that represent where you want the business to go and customers that prevent future success.15
- Treat the first profile as unproven: most founders initially got their ICP wrong.16
- Do not postpone ICP work. Front did not think about it early enough, and Mathilde Collin called that one of her biggest mistakes.17