A useful ICP is not a description of your ideal customer. It is a filter. If someone on your team cannot look at a company and decide within a few minutes whether it belongs on the target list, the ICP is not specific enough yet. Build it from five parts: firmographics you can filter on, a problem you solve well, a trigger that makes the problem urgent, the people involved in the decision, and clear disqualifiers.
Why most ICPs never get used
A typical ICP document says something like: mid-sized companies in manufacturing, healthcare and retail that are investing in digital transformation. It sounds reasonable, but it describes a large part of the market and rules out almost nobody.
When an ICP does not exclude anyone, sales stops using it. Lists get built on gut feel, every account looks like a possible fit, and outbound messages become generic because the list itself is generic. The problem is not effort. It is that the ICP was written to describe, not to decide.
The five parts of a usable ICP
- Firmographics you can actually filter on: industry, company size range, geography and anything visible from the outside, such as hiring activity, technology in use or funding stage. If you cannot find it in LinkedIn or a data tool, it cannot be used to build a list.
- The problem you solve well: one or two problems, written the way the buyer would describe them, not as a list of your features.
- A trigger: the event that makes the problem urgent now. A new sales leader, a funding round, entry into a new market, a legacy system reaching end of support, or rapid hiring in a team you serve.
- The buying group: who feels the problem, who signs off and who can block the decision. In most technology purchases this includes a business owner, a technical evaluator and someone from finance or procurement.
- Disqualifiers: what makes an account a poor fit even when it looks right on paper. Too small to afford implementation, locked into a competitor contract, or no internal owner for the problem.
Disqualifiers are the part most teams skip, and they are often the most useful. They stop the team spending weeks on accounts that were never going to buy.
Start from your best customers, not your total market
Look at the customers you already have. Which ones closed fastest, stayed longest, needed the least convincing and are the kind you would happily take ten more of? What they have in common is the first draft of your ICP.
If you have only a handful of customers, use your pipeline instead. Which conversations moved forward and which went quiet? The pattern in the ones that moved is more reliable than any market sizing exercise.
Test it before you scale it
Ask two people to build a list of 50 accounts from the ICP document without talking to each other. If the lists look very different, the definitions need tightening.
Then run a small, focused outbound test against that list. Look at the quality of the conversations it creates, not only the number of replies. Revisit the ICP every quarter, because it should get sharper as you learn which accounts actually buy.
Check this in your own business
- Can two people on your team build the same target list from your ICP without discussing it?
- Of your last ten closed deals, how many would your current ICP have included?
- Of your last ten lost or stalled deals, would any written disqualifier have flagged them earlier?
- Does each outbound message refer to a problem or trigger that appears in the ICP?
B2B Revenue & GTM Partner for Technology Companies. 10+ years in business growth, GTM, enterprise sales and solution consulting across India, the US, the UK, Europe and the UAE.