The customer who gets the most value and is easiest to reach. If your definition does not tell you where to find them tomorrow, it is not specific enough.
Your Ideal Customer Profile describes the customer who gets the most value from your product, is easiest to reach, easiest to sell to, and most likely to stay.
For B2B it describes an organisation and a role. For consumer it describes a person and a situation.
The test: does it tell you where to find them tomorrow?
Because everything downstream depends on it. Your messaging, your pricing, your channels, your roadmap and your sales process are all different for different customers.
A vague ICP produces vague everything. 'Small businesses' gives you no idea where to advertise, what to say, or what to build next.
Too vague: 'Small businesses that need better scheduling.' You cannot find these people, and the product for a hair salon is nothing like the product for a haulage firm.
Specific enough:
Logistics firms with 20–200 vehicles, in the UK and Ireland, where an operations manager currently builds the weekly rota in a spreadsheet, and where the firm has at least two depots.
Now everything follows. You can buy a list. You know which trade bodies they belong to and which events they attend. You know the job title to search for. You know the product must handle multiple depots. You know the buyer is probably the ops director, not the manager who uses it.
Also note what it excludes. Firms with 5 vehicles do not need this. Firms with 2,000 will want enterprise features you do not have. Saying no to both is what makes the yes work.
First-time founders often resist narrowing because it feels like giving up potential customers. The opposite happens: a narrow ICP makes your messaging sharp, your product decisions obvious and your sales far more efficient. You can widen later from a position of strength.
The second mistake: defining the ICP by firmographics alone — size, industry, geography — with no behavioural trigger. 'Currently builds the rota in a spreadsheet' is worth more than all three of those, because it identifies who has the problem now.
Which interviewees were most desperate? Which paid? Look at what those people have in common.
Size, industry, geography — plus the observable behaviour that means they have the problem today.
Who feels the pain, and who signs. Often two different people. Both belong in the definition.
Could you build a list of 100 of these tomorrow? If not, it is not specific enough to act on.
Explicit exclusions are what make the definition useful. If it excludes nobody, it is not a definition.
Your ICP shifts as you learn. Review it, but do not widen it just because sales are slow.
As soon as you have ten to twenty interviews, and before spending anything on acquisition.
Do not lock an ICP before doing discovery — it will just be your assumptions with a formal name. And do not narrow so far that fewer than a few hundred customers exist.
Apply this to your own startup in My Full Journey (free account).