How to define an ICP for outbound: turn judgment into working rules

An outbound ICP is a set of observable rules for deciding which accounts your company can credibly serve. It combines required characteristics, useful indicators, explicit exclusions and an evidence standard. Its job is to make account selection consistent before research, enrichment and outreach begin.

Many ICP documents describe an aspiration: mid-market technology companies that value innovation and want to grow. That language may help a strategy discussion, but it cannot tell an operator whether a named company belongs in a campaign. Outbound needs a sharper contract between strategy and execution.

Begin with the problem you can solve

When you have them, start from customers that reached value, not from every company that could theoretically buy. Earlier-stage teams can use their best sales and discovery evidence, then mark the assumptions that still need validation. Look for the operating condition that made the problem urgent, the consequence of leaving it unresolved and the people who participated in the decision.

Separate what made the customer a good fit from what merely made the sale possible. A familiar brand, a warm introduction or an unusually determined champion may explain a win without defining a repeatable market.

Write four kinds of criteria

1. Requirements. Conditions that must be true for the account to qualify. These may cover geography, business model, technical environment, customer type or minimum operating complexity.
2. Positive indicators. Characteristics that improve expected fit but are not mandatory. Use them later for prioritization instead of quietly turning every preference into a gate.
3. Exclusions. Conditions that make an account unsuitable, already covered or unsafe to contact. Include existing customers, active opportunities, unsupported markets and business types the offer cannot serve well.
4. Review triggers. Cases where public data cannot support a confident decision. A human review state is more honest than forcing weak evidence into a yes or no.

Make each rule observable

For every criterion, write the definition, acceptable evidence, source preference and missing-data treatment. “Enterprise” is not a rule until the team agrees whether it means revenue, employee count, locations, technical complexity or purchasing structure.

Use examples near the boundary. If the target begins at 200 employees, inspect companies at 150, 200 and 250. If manufacturers qualify but engineering consultancies do not, test businesses that do both. Boundary cases reveal hidden assumptions faster than obvious matches.

Define the account unit

Decide whether the team sells to a parent company, a regional entity, a business unit or an individual site. This choice affects market size, ownership and buying-committee research. Without it, one source may produce a global group while another produces twenty subsidiaries, and both lists can appear correct.

Assign one stable account identifier and keep the parent relationship. The same identity should persist through your TAM map, CRM, outreach and attribution.

Test the ICP against a small market sample

Before sourcing thousands of records, apply the rules to a mixed sample of known customers, lost opportunities, obvious targets and borderline companies. Have two people qualify the same accounts independently. Compare decisions and the evidence used.

Disagreement usually exposes one of three issues: the rule is vague, the required information is unavailable, or commercial judgment has not been made explicit. Revise the contract until a new operator can reproduce the result.

Move from qualification to priority

The ICP decides who belongs in the market. It does not need to decide the exact order of work. Once accounts pass qualification, use an explainable account-tiering model to combine fit, timing, coverage and expected effort.

This separation makes the system easier to change. A new signal can raise an account's priority without rewriting the ICP. A new exclusion can remove accounts before the team spends money enriching people who should never enter the motion.

Treat the ICP as a governed decision

Name the person who can approve a rule change, record why it changed and measure the downstream result. Review rejected accounts, qualification rates and opportunity outcomes by criterion. The goal is not a permanently fixed profile. It is a market definition that gets clearer as evidence accumulates.

A working ICP should leave the team with fewer arguments at the list stage and better questions after launch. When a play underperforms, operators can inspect whether the accounts matched, whether the timing was relevant and whether the buying roles were covered, instead of blaming one undifferentiated list.

Use the free ICP Qualification Rule Builder to turn those decisions into an operator-ready contract.