Ask five founders for their target customer and four will say some version of "B2B companies that need X". That is not a target; it is a direction. An ideal customer profile turns the direction into a filter you can actually run a business through: who to build for, who to write for, and exactly who to put in the next campaign.
What an ICP is, and what it is not
An ideal customer profile describes the company most likely to buy quickly, succeed with the product, and renew. It is written in observable, filterable attributes: industry, headcount, geography, business model, tooling, and the trigger event that makes the problem urgent.
It is not a buyer persona. The persona is the person inside the ICP company: the champion who feels the pain, the executive who signs, the ops lead who can veto. You need both, in that order. Personas without an ICP are fiction with a job title.
Five ICP examples you can steal the shape of
1. Dev-tool startup selling an API monitoring product. B2B software companies, 20 to 200 engineers, running microservices on AWS or GCP, already paying for Datadog or Grafana Cloud, whose job postings mention reliability or SRE. Trigger: a public status-page incident in the last quarter.
2. Agency-replacement SaaS (like Cafiyn FlyWheel's own ICP). Founder-led B2B software companies, 1 to 20 people, product live and priced, no sales team, buyer is a business decision-maker. Trigger: launch traffic faded and the founder is doing outbound by hand.
3. Compliance automation for fintech. Seed to Series B fintechs in the US and EU, 10 to 100 employees, handling regulated data, no in-house compliance hire. Trigger: first enterprise deal stuck in security review, or first audit scheduled.
4. Ops platform for logistics. Regional freight and 3PL operators, 50 to 500 employees, dispatch still coordinated over spreadsheets and phone calls, low software maturity. Trigger: a new ops director hired from a larger, better-tooled competitor.
5. AI writing tool for agencies. Marketing agencies, 5 to 50 seats, retainer-based, producing content for 10+ clients monthly, already paying for two or more point tools. Trigger: margin pressure named in their own positioning ("we do more with less").
Notice what every example shares: each attribute is queryable in a database, a funding tracker, or a LinkedIn filter, and each ends with a trigger. Fit tells you who could buy; the trigger tells you who will buy now.
The template
Fill these eight fields and you have an ICP precise enough to campaign against:
- Industry / category: narrow enough that the same message lands on everyone in it.
- Company size: a range, in employees or revenue, that matches your price and sales motion.
- Geography: where you can legally sell, support, and be paid.
- Business model: B2B or B2C, self-serve or sales-led, whatever changes how they buy.
- Tooling / stack: what they already pay for (your integration point or your replacement target).
- Trigger event: the observable change that makes the problem urgent this quarter.
- Buying committee: champion, signer, blocker, by role title.
- Disqualifiers: attributes that predict churn or a bad sale, written down so nobody chases them.
How to validate the profile
An ICP is a hypothesis until a list and a market agree with it. Two tests: first, build a list of 50 real companies that match every field; if you cannot, the profile is too narrow or too vague. Second, check the sizing math holds: run the profile through the bottom-up method in our TAM, SAM, SOM guide and confirm the obtainable slice supports your next milestone. The broader validation playbook, including interviews and willingness-to-pay, is in how to validate a startup idea.
This is also exactly the work Cafiyn Lens automates: its ICP and buyer discovery module produces the profile, the buying committee, and the evidence behind both, and the same profile then drives FlyWheel campaigns without being re-entered anywhere. One profile, written once, campaigned against directly: that loop is the point of a go-to-market engine.