How to Build an ICP That Actually Filters

An ideal customer profile is a definition of the company most likely to buy, stay and be profitable. A useful ICP removes companies from consideration in under ten seconds. Build it from your last twenty closed deals using four layers: firmographics, technographics, buying triggers and an explicit negative filter.

Most ideal customer profiles are useless. Not wrong exactly, just too broad to make a single decision. “B2B SaaS companies in Europe with 10 to 200 employees” describes roughly forty thousand businesses and tells a campaign nothing about which of them to contact first.

A real ideal customer profile does one job: it removes companies. If your ICP does not let you look at an account and say no in under ten seconds, it is not an ICP. It is a description of your market.

Start with evidence, not aspiration

The most common mistake in building an ICP is starting with who you want to sell to. Start instead with who has already bought, stayed, and been profitable.

Take your last twenty to thirty closed deals and separate them into three groups: the ones that closed fast and are still customers, the ones that closed but churned or were painful, and the ones that never closed despite looking promising.

The pattern lives in the difference between group one and group three. That is your ICP.

Layer one: firmographics

The basic filters. Necessary but not sufficient, and where most ICPs stop.

• Industry, specific enough to be actionable. Not “technology” but “logistics software vendors selling to freight forwarders”.

• Company size, in employees and revenue. Both, because they diverge in interesting ways.

• Geography, at the level you actually operate. Country, and often region within country.

• Growth stage, funding status, recent raise, ownership structure.

• Business model, since a subscription business and a project business buy differently even in the same sector.

Layer two: technographics

What a company already runs tells you more about buying readiness than its size does. A company using Pipedrive and an email sequencing tool has already decided outbound matters. A company with no CRM has a different problem and a much longer sales cycle.

• CRM in use, which signals sales maturity.

• Marketing automation, which signals budget and process.

• Competing or adjacent tools, which signal both budget and dissatisfaction.

• Absence of a category of tool, which can be the strongest signal of all.

Layer three: the trigger

Firmographics tell you who could buy. Triggers tell you who might buy this quarter. This is the layer that separates campaigns that book meetings from campaigns that get ignored.

• Hiring signals. A company posting for two SDRs is investing in outbound and has just admitted the problem publicly.

• Funding. A recent raise means a growth target and a deadline.

• Leadership change. A new CRO or head of sales in the first 90 days is actively looking for things to change.

• Market expansion. A new office, a translated site or a job posting in a new country.

• Product launches that need pipeline behind them.

• Competitor switching signals, visible in job posts or public complaints.

Layer four: the negative filter

The part almost everyone skips, and the part that saves the most money. Write down explicitly who you will not contact.

• Below a revenue floor where the deal cannot be profitable.

• Industries with procurement cycles you cannot survive.

• Companies without the internal capacity to use what you sell.

• Segments where your last four deals all churned.

• Geographies you cannot service in the local language or time zone.

A negative filter is what makes an ICP operational. It is the difference between a list of 40,000 and a list of 900 that a human could work through deliberately.

The buying committee, not the buyer

B2B purchases at any meaningful value involve several people. Your ICP should name the roles, not just the company.

• The champion, who feels the pain daily and will advocate internally.

• The economic buyer, who releases budget and often never replies to cold email.

• The blocker, usually finance, legal, IT or procurement.

• The end user, whose resistance quietly kills adoption after signature.

Cold outreach usually targets the champion, but the message has to survive being forwarded to the economic buyer. Write for both.

Testing whether your ICP is real

Three checks. If it fails any of them, it needs another pass.

• The ten second test. Show a colleague ten company names. If they cannot sort them into yes and no using your ICP alone, it is too vague.

• The list size test. Your ICP should produce a list you could realistically work in a quarter. If it produces 40,000 accounts, add filters until it does not.

• The message test. If you cannot write a specific opening line that would only make sense to companies in your ICP, the profile is not specific enough to be useful.

Review it quarterly. An ICP is a hypothesis, not a document. Every closed deal and every lost deal is evidence. The companies you are winning most easily in month nine are rarely the ones you wrote down in month one, and the profile should follow the evidence rather than the original assumption.

Hear from the 50+ clients we have built ICPs for, or pressure test your ICP with us.

Frequently asked questions

What is an ideal customer profile?

An ICP defines the company most likely to buy, stay and be profitable. It differs from a buyer persona, which describes an individual. A working ICP is specific enough that a colleague could sort ten company names into yes and no using it alone.

How is an ICP different from a buyer persona?

An ICP describes the company: industry, size, geography, technology, growth stage. A persona describes the person inside it: role, priorities, objections. You need both. The ICP decides which accounts to contact, the persona decides who within them and what to say.

How many companies should my ICP produce?

A list you could realistically work through in a quarter. If your filters return 40,000 accounts, keep adding criteria until they do not. Precision is what makes outbound affordable, because relevant email gets replies and replies protect sender reputation.

What are buying triggers?

Events suggesting a company might buy now rather than eventually: a recent funding round, two SDR job postings, a new CRO in their first 90 days, market expansion, a product launch, or public dissatisfaction with a competitor. Triggers separate campaigns that book meetings from campaigns ignored.

How often should I update my ICP?

Quarterly. An ICP is a hypothesis, and every closed and lost deal is evidence. The companies you win most easily in month nine are rarely the ones written down in month one. Follow the data rather than the original assumption.