Most ideal customer profiles describe an audience: industry, company size, job title, and a stock-photo persona with an alliterative name. They get built once, presented once, and never opened again. A useful ICP is different. You should be able to read it and know what to change in your ad account that afternoon.
Demographics describe, behaviour predicts
A job title tells you who somebody is. It does not tell you whether they will buy. Two people with identical titles at identically sized companies can have completely different odds of converting, because one of them just lost a key supplier and the other is perfectly content.
Behavioural signals predict. What have they already tried and abandoned? What adjacent thing have they already paid for? What changed recently in their business? Those tell you about intent. Demographics only tell you about eligibility.
Start from the customers you already have
Not the ones you wish you had. Open your closed deals and sort them honestly.
Which ones closed fastest? Which stayed longest? Which needed the least hand-holding and delivered the best margin? Which ones came back and bought again without being chased?
The pattern in those answers is your ideal customer profile, whether or not it matches the market you told investors you were going after. This is uncomfortable often enough that plenty of teams skip it.
Your ICP is a finding, not a decision. It already exists in your closed-won data.
The four things worth writing down
Everything else is padding. These four each map to something you can act on:
- The trigger. What changed that made them start looking, and roughly when in their year it happens.
- The alternative. What they would have done instead, including doing nothing.
- The objection. The thing that nearly stopped them signing.
- The disqualifier. What makes somebody a bad fit, stated plainly enough to act on.
Exclusions are where the money is saved
Almost every ICP document says who to target. The version that changes your spend also says who to actively avoid, and that half is usually missing.
Write down the segments that convert cheaply and churn immediately. The ones that need three times the support for the same fee. The ones that never reach a decision no matter how many follow-ups they absorb. These are not theoretical — you already know who they are, because your team complains about them.
A cheap lead that costs you four hours of unbillable scoping is not a cheap lead. Cost per acquisition does not capture that, so your ICP has to.
Turning it into actual spend decisions
Here is the test of whether the document is real. Each line should map to a setting somebody can change:
- Trigger becomes keyword selection and campaign timing.
- Alternative becomes your ad copy and your comparison messaging.
- Objection becomes a section on the landing page and a line in the follow-up email.
- Disqualifier becomes negative keywords, audience exclusions and a qualifying question on the form.
Run the document against that mapping. If your ICP has not caused a single targeting change, a single new negative keyword, or a single rewritten page, then it is not an ideal customer profile. It is a description, and descriptions do not compound.
