What Is a Brand ICP? Start With Who You Are Not For

A brand ICP, or ideal customer profile, is the narrow slice of the market you are built to win. The fastest way to sharpen yours is to run it backwards and name who your brand is not for.

Years ago I sat in a go-to-market meeting with a national B2B platform company. They had built the infrastructure under well-known consumer brands and wanted to launch one of their own. The chief revenue officer wanted to figure out which customers to chase. I told him to start with who it wasn't. He nodded, said that made everything easier, and he was right.

The refusal list is where conviction lives.

What an ideal customer profile actually is

ICP stands for ideal customer profile, and all three words carry weight. Ideal points at one customer: the one who gets the most from what you sell, pays with the least friction, stays longest, and lifts the business by being in it. Customer means the specific person, household, or company you close, serve, and keep. Profile means attributes you write down and act on. A profile is a pattern.

Most founders think they hold an ICP when they hold a target market: a broad category like enterprise SaaS or cannabis. An ICP is the narrow cut of that category you plan to own. The Fortune 500 logo you wish you had stays fan fiction until you are closing those deals.

The Five Filters that narrow a brand ICP

Most ICP work stops at three lenses: demographics, psychographics, behavior. Three is enough to launch a brand, and most agencies stop there. I run five, built over years of engagements across cannabis, healthcare, real estate, and B2B. Each screens tighter than the last, and each carries its inverse: who this is not for. The inverse is the half that does the work.

• Category. The shelf your customer files you under. For Knack, cannabis; for Liquid Death, bottled water; for your firm, whatever someone Googles when the problem hits.
• Geography or context. Where they live, work, or buy. Knack chose New York; Liquid Death chose music venues and tattoo parlors long before grocery aisles.
• Lifestage or revenue band. Age and household for consumer, revenue and headcount for B2B. A fifty-person startup's CRO and a twenty-thousand-person enterprise's CRO are different animals.
• Trigger. What sends them looking. "I'm bored with what I have." "Our last vendor failed." Most documents skip it, and it is the one teams miss most.
• Psychographic. Who they believe they are. Knack's customer sees a New Yorker with taste; Liquid Death's sees someone who hates plastic and refuses to be sold to. Self-image runs deepest, because you must hear what people never say out loud.

By the fifth filter, that crowd has shrunk to a pattern tight enough to run a brand against.

Where the inputs come from

A framework does nothing without inputs. Pull them from three places.

• Your ten best customers. Interview the ones you would clone. Ask what sent them looking, what they almost bought instead, and what words they use to describe you. That last answer hands you the language your future customers already speak, which is April Dunford's point in Obviously Awesome.
• The unfiltered record. Sales transcripts, support tickets, reviews. The way customers talk when they are not flattering you is the most accurate ICP data you will ever hold, and most companies leave hundreds of hours of it unread in a CRM.
• Adjacent purchases. What people buy alongside you reveals the category they actually file you under, sometimes a surprise.

Fill in the five filters by hand and write the inverse of each. If the "not" side is empty or comfortable, you have not committed, because that side is revenue you are choosing to wave off. Read it in your next leadership meeting: whoever flinches marks where your budget is bleeding.

Knack: the profile sharpened in two passes

My studio led brand and strategy for Knack, the first recreationally grown cannabis brand in New York, grown in the Adirondacks and sold in state dispensaries. Its target market is New York recreational cannabis, roughly two million people. Its ICP is far narrower, and the narrowing gave the brand a voice.

We ran it in two passes. The first was directional: local, legal, built on volume, fair prices, and real product, clear of both the stoner cliché and the luxury labels. Enough to start naming, and we landed on Knack. Four letters, percussive, ownable. The word assumes an adult with taste, a knack for something, rather than someone chasing the strongest thing on the shelf.

Then the name pulled the profile tighter, so we ran a second pass: a New Yorker with identity-level pride who buys cannabis the way they grab a six-pack. That pass made the copy possible: "another reason for Jersey to be jealous," "once you go Knack, you don't go back." ICP moves with the brand. Do it badly and the document goes in a drawer.

Liquid Death: same discipline, different shape

Start with the name. Every other brand in bottled water signals purity or wellness: Aquafina, Smartwater, Pure Life, Essentia. Liquid Death named itself the opposite of all of them, and that was the ICP decision spoken out loud in two words on the can.

It is canned water in a tallboy that reads like a beer, sold in tattoo parlors and music venues before grocery. Mike Cessario, a former Netflix creative director, founded it in 2017; by 2024 it was valued near one and a half billion dollars. The idea traces to the 2009 Vans Warped Tour, where he watched concertgoers drink water from Monster Energy cans. A Monster can did not make you feel like a kid, and a plastic bottle did.

So the psychographic runs in two layers: people who refuse to keep buying plastic, and people who feel uncool holding water in public. Knack runs on New York pride; Liquid Death runs on skulls and death-metal type. Same structure, opposite shapes, because the profile wrote each playbook.

Both brands walked away from someone. Specifically, on purpose, in writing.

Where ICP work fails, and why it pays off

Skip the profile and the math turns against you. Acquisition cost climbs, conversion falls, retention thins, and the brand sounds like everyone else because it addresses everyone. Run a real ICP and the math reverses: cheaper acquisition, higher conversion, compounding retention, pricing power.

The work fails in four ways. It stays too broad and never narrows past the target market. It chases whatever looks closeable this quarter, when the profile is a multi-year commitment. It calcifies onto a segment the market has already left. Or it slides into founder affinity, when the real question is who generates the most value, pays reliably, and grows with you.

This was never a marketing exercise. It is a financial decision that happens to show up in the marketing.

The takeaway

Run the Five Filters on your own business. Do the trigger work, the psychographic work, and the inverse of all five. The "not" half is where the conviction lives.

If you cannot describe your ICP in one paragraph, with real attributes, a real trigger, and a specific list of who you are not for, writing that paragraph beats a rebrand or a new website for value this quarter. If you already have the document, the only question is whether the profile shows up in the work: on the packaging, in the voice, on the page that lists what you will never say.

If it doesn't, the document is decoration.

This is the written version of Position to Win, Episode 3.