
STRATEGY —
You Don't Have an ICP
Most "ideal customer profiles" just describe everyone on the planet.
TL;DR: We recently worked with a client that had already done what most companies skip: they built an ideal customer profile. Look closer, though, and it described nearly everyone on the planet, avatar after avatar until the whole map was covered. The missing piece wasn't more research, it was the word ideal itself, earned only by isolating the customers who are the most profitable, cheapest to acquire, longest-tenured, and highest lifetime value. My partner Ryan Deiss adds the counterintuitive payoff: narrowing the aim reaches more people, not fewer, because a tight target still catches its near-misses.

The Profile That Wasn't
A client of ours recently handed over their ICP, and I want to give them credit for that first. Most companies never build one. They run on inherited assumptions about who buys and why, never written down, never tested. This company had done the actual exercise: personas built the way marketing school tells you to build them first.
Then we looked closer. It was effectively everyone on the planet.
How the Aperture Widens
That's not a sign of laziness. It's what happens by default. You build a profile for the segment you started with, the business grows, you find you can serve the segment next door too, so you add a persona for that one as well. Each addition makes sense on its own.
My partner Ryan Deiss calls it widening the aperture: it happens gradually as a company scales, until you're technically targeting everybody, which functionally means you're targeting nobody. Acquisition costs climb. Conversion drifts down. And because the targeting looks thorough on paper, avatar after avatar, full coverage, it's the last thing anyone thinks to question.
The Letter That Was Never Earned
Here's the distinction that got missed, and it's sitting inside the acronym everyone already uses. ICP stands for ideal customer profile. What this company had built was a customer profile. The ideal part was never earned. It was assumed.
Earning it isn't a branding exercise, it's an analytical one. Take the actual customer data, hand it to finance or a data scientist if you have one, and ask a narrower question than "who buys from us."
Ask who is the most profitable. Who costs the least to acquire. Who stays the longest. Who carries the highest lifetime value. That intersection, not the union of every persona you've ever written, is the ideal. This company already had the history and the data. What it had never done was run that specific cut.
Be honest: is your ICP built from data, or from guesswork?
Why Narrowing Reaches More People
The part that stops most owners from doing it is the part that sounds obvious once someone else says it: narrowing the target feels like giving something up. If you've spent years telling yourself the addressable market is everyone, shrinking it down to the profile that actually performs reads as retreat.
Ryan's framing of why it isn't is the part worth keeping. Think of it as a bullseye. Aim for the center and miss slightly, and you still land on people in the outer rings, not the ideal, but close enough to be drawn in anyway. Aim at everyone, and you're not even on the target. Narrow the aim and you end up reaching more people, not fewer, because a tight target still catches its near-misses. A blanket doesn't catch anything; it just covers everything equally badly.
The first move here isn't a new campaign or a bigger budget. It's going back through data the company already owns and asking a harder question than the one marketing school teaches you to ask first, then being willing not to like the answer if it shrinks the map.
— Roland

Want more than just the weekly deep dives?
On Instagram we share quick tips, behind-the-scenes looks, and first access to what's coming next.
Follow @RolandFrasier on Instagram and join the community.
Thinking About Exiting Your Business?
What Would Buyers See If They Evaluated Your Business Today?
Your Exit-Ready Score reveals the hidden risks that suppress valuation, built on the same indicators private equity uses to screen deals in under 5 minutes.
Find out your score….

Roland's Riff
I stopped testing my app on real people once I realized they were working around the issues instead of identifying them.
Real users adapt. They route around the broken thing, and the broken thing stays invisible.
So I built ten AI personas and gave each one an actual job to do inside it: sign up as a new customer, get through onboarding.
None of them made it in. All ten failed at the exact same step, the front door.
That is the difference between describing a customer and giving one a job. A description tells you nothing. A result tells you who you are actually built for.
Want the full breakdown? Read the post.



