How to Define Your ICP (With Examples)

“Anyone with a pulse and a credit card” is not an Ideal Customer Profile. Neither is “B2B SaaS companies” or “agencies in North America.” Those are markets, not ICPs, and confusing the two is the most common reason cold outbound campaigns underperform. An ICP is a tight, specific, actionable description of the exact kind of company that gets the most value from your offer in the shortest time with the least friction. The narrower it is, the better your campaigns will perform — which sounds counterintuitive but is one of the most consistent patterns we see across hundreds of clients.

This guide walks through the framework we use during onboarding with every new ReplyWorks engagement. It takes about ninety minutes to do well and saves months of wasted sending.

Why narrow beats broad — every single time

Most founders we work with come in convinced their ICP is wide. “We could sell to anyone,” they say, and they are usually correct in a strict sense. Their product technically works for many kinds of buyers. But “could sell to” and “should target” are very different questions. When you target broadly, your messaging has to be generic enough to appeal to everyone, which means it lands powerfully on no one. When you target narrowly, you can write copy that names the specific pain, the specific tools, the specific job title, and the specific consequence — and that kind of specificity is what gets reply rates above five percent.

Narrow targeting also dramatically simplifies every other part of your outbound system: data sourcing becomes easier, lookalike modeling becomes possible, your case studies start to compound because they all sound alike, and your sales conversations become repeatable because every prospect arrives with the same problems. Broad targeting compounds nothing.

The five-dimension ICP framework

A useful ICP is defined along five dimensions. Skip any one of them and you have a market segment, not an ICP.

1. Firmographics

Industry (be specific — not “SaaS” but “vertical SaaS for the construction industry”), employee count range (not “SMB” but “twenty-five to seventy-five employees”), revenue range, geography, and funding stage. The tighter the bands here, the more useful the ICP. “$5M–$20M ARR” is usable. “SMB to enterprise” is not.

2. Technographics

What tools do they already use? This is one of the most underutilized targeting dimensions in B2B. If your product integrates with HubSpot but not Salesforce, that is not a footnote — that is a hard constraint that should appear at the top of your ICP. Tools like BuiltWith, Wappalyzer, and Clay’s integrations let you filter prospect lists by exact tech stack, and a “uses HubSpot + Stripe + Webflow” filter is dramatically more powerful than a generic industry filter.

3. Trigger events

What happened recently in this company that makes them more likely to need you right now? New funding round, new VP of Sales, recent layoffs, opening a new office, hiring spree in a specific function, recent compliance changes, recent acquisition, new product launch. Trigger events are the single highest-leverage targeting signal in cold outbound because they explain why the prospect should care this week rather than someday.

4. Pain signals

What observable evidence suggests this company has the problem you solve? Job postings mentioning the pain, public reviews complaining about it, a competitor they recently churned from, a technology they use that signals the problem (e.g., still using Mailchimp for outbound), a recent blog post from their leadership describing the issue. These are inferable from public data and are the difference between guessing and knowing.

5. Decision-maker profile

Who actually buys, and what is true about them? Title, seniority, tenure (people in the first ninety days of a new role buy more), department, reporting line, and the specific goal they are accountable for this quarter. “VP of Sales” is too broad. “VP of Sales at a Series-A SaaS company in their first six months on the job, accountable for outbound pipeline” is an ICP buyer profile.

A worked example

Here is a real ICP from a recent client engagement, with the company name removed. The product is an AI-powered QA tool for software teams.

Bad ICP (what they walked in with): “B2B SaaS companies that need better testing.”

Good ICP (what we built together): “Series A or Series B B2B SaaS companies headquartered in the United States, fifty to one hundred fifty employees, with at least fifteen engineers, currently using Cypress or Playwright but not yet a paid AI testing tool, in the past sixty days have either shipped a major release that introduced regressions (visible from changelog or social) or posted a job opening for a QA engineer or SDET. The buyer is the VP of Engineering or Director of Engineering, in the role for less than eighteen months, and accountable for either release velocity or production incident rate.”

That is a long sentence. It is also targetable — every condition in it is filterable using Clay, Apollo, Sales Navigator, and a couple of public data sources. The total addressable list this ICP produced was around eight hundred companies, not eighty thousand. And those eight hundred converted to meetings at four times the rate of the previous broad list of “B2B SaaS engineering leaders.”

The skip list — who NOT to target

Equally important and often skipped: write down who is explicitly out of scope. Companies under a certain size. Industries you cannot serve. Geographies your product does not support. Decision-makers who do not have authority. Tech stacks that are incompatible. Stages of growth where the pain has not yet appeared. Past customers who churned for fixable reasons — and past customers who churned for unfixable reasons (skip those forever).

The skip list is not a rejection. It is a focusing tool. Every company you remove from the target list is a company you do not waste an email on, do not send a follow-up to, and do not pretend to qualify on a discovery call. That preserved energy compounds into the prospects you should be talking to.

How to actually do this exercise

Block ninety minutes. Pull up your last twenty closed-won customers and your last ten closed-lost or churned customers. For each one, fill in the five dimensions above. Look for the patterns that show up in your wins and the patterns that show up in your losses. The overlap of “appears in wins” and “absent in losses” is your real ICP — not the one you wish you had, the one your data already proves works.

Then go to your sales team and have them reality-check it. Then go to your top three customers and ask them what their world looked like the week before they bought from you. Combine those three inputs, write a single paragraph, and that is your v1 ICP. Revise it every quarter as more data accumulates.

This is unglamorous work. It is also the highest-leverage hour of strategy you will spend on outbound this year.