How to Find Your Ideal Customer Profile (ICP) in 2026

Published 2026-06-01 · 10 min read · 2,400 words

Most first-time founders skip Ideal Customer Profile work and burn months building for "everyone." This guide shows you the 60-minute manual sprint and the AI-assisted shortcut using ICPGTM.

Why most first-time founders skip ICP work (and pay for it)

Almost every founder I meet can tell me what their product does in two sentences. Very few can tell me, with conviction, *exactly* who is supposed to buy it.

"Small businesses." "Creators." "Students." "Anyone who wants to get fitter."

These are not customers. They are crowds.

If you cannot describe your buyer in enough detail that you could walk into a coffee shop and physically point one out, you do not yet have an Ideal Customer Profile — and every dollar you spend on ads, every cold email you send, and every feature you build is a guess wrapped in hope.

This guide shows you two ways to fix that. A 60-minute manual sprint you can run today, and an AI-assisted shortcut using a tool called ICPGTM that compresses days of research into about 90 seconds.

> Heads up: ICPGTM is a tool I built. Every Expansary course participant gets 25 free credits to use on it — enough to generate several full playbooks while you work through the course.

What an ICP actually is

An Ideal Customer Profile has four ingredients. Miss one and the whole thing wobbles.

1. Company shape — size, industry, stage, business model, geography, tech stack. The container your buyer sits inside. 2. Buyer persona — role, seniority, what they are measured on, what makes their week harder. 3. Acute pain — the specific problem your product removes. Not "wants to grow." Acute. "Loses 6 hours a week chasing late invoices." 4. Buying trigger — the event that turns "this would be nice" into "we need this now." A new hire, a funding round, a churn spike, a regulation change.

A demographic ("women aged 25–34") is not an ICP. A job title ("Head of Marketing") is not an ICP. A vibe ("ambitious creators") is *definitely* not an ICP.

An ICP sounds like this:

> *Seed-stage B2B SaaS companies (5–25 people, $0–$2M ARR) where the founder is still running sales themselves, has just hired their first AE, and is bleeding deals because nobody is qualifying inbound leads.*

That sentence is targetable. You can list 200 companies that match it by lunchtime.

The 4 inputs you need before you start

Before you generate, sprint, or interview anything, get these four things on paper:

• A one-line product description. What it does, for whom, why it is better. 25 words max.
• Your current best guess at the buyer. Even if it is wrong. You need a hypothesis to test.
• A "why now" story. What has changed in the world that makes this product needed *this year*?
• Any traction signals you already have. Even three friendly users count. Especially three friendly users.

Got them? Good. Now you have two paths.

Path 1 — The 60-minute manual ICP sprint

If you have zero budget and one free hour, do this:

1. List everyone who has ever paid you, signed up, or said "I'd buy that" (10 min). Even free users count. 2. Tag each one with 4 attributes: company size, industry, role of the person, and the specific pain that made them say yes (15 min). 3. Look for the cluster. Almost always 60–80% of your "yes" signals come from a tighter slice than you expected (10 min). 4. Write the cluster up as a single ICP sentence using the template above (10 min). 5. List 20 companies that match it using LinkedIn search, Crunchbase, or just Google. If you cannot find 20, your ICP is too narrow. If you find 2,000 in five minutes, it is too broad. Aim for "I had to think a bit" (15 min).

You now have a working ICP. It is not perfect. It is enough to start outreach.

Path 2 — Generate a full playbook with ICPGTM

The manual sprint works. It is also a lot of typing. When I am pressure-testing a new idea, I open ICPGTM and let it draft the entire playbook for me, then I edit.

Here is the flow, end to end:

Step 1 — Describe your product (about 60 seconds)

You give it a name, a one-line description of what it does, and your URL if you have one. It pulls extra context from your site automatically, so you do not have to over-explain.

Step 2 — Generate the playbook (about 90 seconds)

It thinks, and ships six artefacts at once:

• 3 ranked ICP profiles — fit-scored company shapes, buyer personas, pain points, budget ranges, and buying triggers
• A buyer committee for each ICP — decision-maker, champion, blocker, influencer (this alone usually surfaces a person I had forgotten to message)
• Outreach playbooks — cold email templates, LinkedIn DMs, elevator pitches, plus which channel each ICP actually pays attention to
• A personalized message for one specific target — paste a real company URL and it tailors a cold email + LinkedIn DM to them
• A full GTM plan — positioning, motion (PLG vs sales-led), pricing notes, a 30/60/90 rollout, content ideas, objection handling
• Version compare — regenerate any section with feedback, diff it against the previous draft, and compare two ICPs side-by-side without losing prior work

Step 3 — Review the three ranked ICPs

This is where you, the human, earn your keep. The model is good. It is not you. Read each profile and ask:

• Have I actually met someone like this in the last 6 months?
• Can I name 10 companies that match this shape right now?
• Is the "buying trigger" something that actually happens often, or once a decade?

Step 4 — Personalize for one real company

Pick a target you genuinely want as a customer. Paste their URL. ICPGTM tailors the outreach to them specifically, saved under a dedicated Personalized tab. This is the unlock — generic templates get ignored, but a 4-sentence cold email that references the prospect's last funding round and product launch gets replies.

Step 5 — Refine and re-run

Hit regenerate on any section with feedback like "make ICP 2 more enterprise-leaning" or "rewrite the cold email more direct, less consultative." Diff the versions. Keep the best lines.

How to pick which of the 3 ICPs to chase first

Founders almost always pick the ICP with the highest fit score. That is usually wrong.

The right ICP to start with is the one that wins on three axes at once:

• Fit — your product genuinely solves their acute pain
• Reachability — you can actually get in front of them this week, not in theory
• Urgency — their pain is on fire *right now*, not "next quarter when we have headcount"

A 92/100 fit score for "Fortune 500 CIOs" is worthless when your only way to reach them is a $4,000 conference ticket. A 78/100 fit score for "indie SaaS founders in a Slack community you are already in" is gold.

Pick reachable urgency every time.

Turning the playbook into your first 10 cold outreaches

A playbook that lives in a tab is worth nothing. Once you have your ICP and your messaging, the next move is mechanical: list 20 companies that match, message all 20, learn from the responses (and the silences), iterate.

We wrote a full walkthrough of that exact process here: How to get your first 10 customers (from zero).

Common ICP mistakes

• Too broad. "SMBs" is a country, not a customer. If your ICP description fits more than 100,000 companies, tighten it.
• Confusing persona with ICP. "Marketing managers" is a persona. The ICP is the *company* the marketing manager works at — size, stage, industry, trigger.
• Ignoring the buyer committee. B2B deals die because of the people you did *not* message — the security reviewer, the finance gatekeeper, the skeptical peer. The buyer committee view in ICPGTM exists to stop you walking past them.
• Setting it and forgetting it. Your ICP should shift every time your product, pricing, or positioning shifts. Re-run it quarterly at minimum.

Bottom line

You do not need a perfect ICP on day one. You need a *specific* one — specific enough that you can name 20 companies that match it, name the person inside who feels the pain, and name the trigger that makes this month the month they buy.

Do the manual 60-minute sprint to build your own conviction. Then use ICPGTM to pressure-test it, generate the outreach assets, and personalize for the real companies on your list.

And if you are taking the Expansary course — those 25 free credits are sitting there waiting. Use them on three different versions of your idea before you commit to one. That is exactly the kind of cheap iteration this whole course is built around.