How to Define Your ICP When Everyone Is a 'Good Fit'

The most common ICP I see at growth-stage companies is "B2B companies with 10 to 500 employees that have a sales team." That's not an ICP. That's a description of the addressable market. An ICP is a specific profile of the customer who closes fastest, pays most, churns least, and refers most — and the factors that predict whether a prospect will become that customer.

The reason vague ICPs persist is that they feel safe. If everyone is a good fit, you never have to turn anyone away. But vague ICPs produce vague marketing, inefficient sales, and reps who spend time on deals that were never going to close. The cost is invisible until you do the analysis.

What an ICP Actually Is (and Isn't)

An ICP is a description of the company and situation that produces your best customers — not the customers you want to have, and not the customers you have the most of. Your best customers close quickly, pay on time, don't churn, expand over time, and refer other customers. Those are the customers your ICP should describe.

An ICP is not a persona. A persona is a description of a buyer — the job title, the goals, the fears. An ICP is a description of the company — the size, the industry, the growth stage, the tech stack, the trigger event that makes them ready to buy. Both matter. The ICP gets you to the right companies. The persona gets you to the right person inside them.

The Problem With "We Serve Any $5M–$50M B2B Company"

When your ICP is this broad, your marketing can't be targeted because there's no signal to target to. Your reps can't qualify efficiently because there are no disqualifiers. Your win rate data looks noisy because you're lumping together deals that had nothing in common except a generic size range.

The practical consequence: you're spending equal sales and marketing resources on companies that close in 30 days and companies that close in 180 days — or don't close at all — and you can't tell the difference in advance. That's expensive.

How to Build an ICP From Your Own Data

Pull Your Last 20 Closed-Won Deals

Export your last 20 to 30 closed-won deals from your CRM. For each one, note: company industry, company size (employees and revenue if available), deal size, sales cycle length, lead source, and who the champion was. If you can also pull your 10 worst customers — highest churn, lowest expansion, most support tickets — do that too.

Find the Patterns

Look for clusters. Which industry shows up most in your best deals? What size range closes fastest? Which lead source has the highest close rate? Which role in the organization is most often the champion? Is there a trigger event — new funding, new leadership, a specific regulatory change — that shows up before the deals that closed fastest?

You're looking for two or three factors that, when they're all present, predict a fast, high-value close. Those factors become your ICP.

Write the Negative ICP

This is the step most companies skip. The negative ICP defines who you shouldn't take — the profile that looks like a fit on the surface but consistently underperforms. Usually it's a specific size range that's too small, a specific industry where your product doesn't quite fit, or a specific situation (too early stage, too price-sensitive, looking for a vendor rather than a partner) that produces bad outcomes.

Writing the negative ICP is important because it gives your team permission to disqualify. Without it, reps default to optimism — "maybe this one will be different" — and your pipeline fills with deals that are unlikely to close.

How to Turn Your ICP Into CRM Routing Logic

Once you have a specific ICP, build it into your CRM as qualifying fields. Industry, company size, lead source, and presence of a trigger event should all be required at the first stage gate. Build a lead score that weights these factors and routes high-ICP leads to your best reps first. Use routing automation to enforce the prioritization consistently.

For companies using AI lead scoring, a clean ICP definition is the prerequisite. You can't train a model on "good fit" — you need specific, measurable attributes. The ICP work comes first; the AI layer comes second.

How to Know When Your ICP Is Wrong

Your ICP is wrong if your win rate at ICP-qualified deals isn't meaningfully higher than your overall win rate. It's also wrong if your reps are regularly arguing that a deal "feels right" but doesn't match the ICP criteria — that's a signal the criteria are missing something real. Review your ICP every six months against new closed-won and churned data. It's a living document, not a one-time exercise.