Ideal Customer Profile

What is Ideal Customer Profile?

An Ideal Customer Profile (ICP) is the firmographic description of the company that gets the most value from your product and is worth building a repeatable go-to-market motion around. It answers one question at the organization level: which companies should we even be talking to.

HubSpot draws the line clearly: "An ideal customer profile (ICP) defines the perfect company for your product or service," built from firmographic, technographic, and behavioral attributes like industry, company size, tech stack, and growth stage (HubSpot). That's distinct from a buyer persona, which describes the individual inside that company rather than the company itself. As HubSpot puts it, "Personas tell you who you're speaking to. ICPs tell you which companies are worth speaking to in the first place." A persona covers a person's role and how they like to be communicated with. An ICP never mentions a single person. It describes the company: headcount, industry, stack, funding stage, the shape of the problem they have.

There isn't much dispute about the term's modern usage in B2B SaaS. Where founders disagree is on how early to nail it down. In Lenny Rachitsky's guide to product-market fit, Census co-founder and CEO Boris Jabes reflects on skipping the exercise early on: "I didn't think about ICPs whatsoever. However, looking back, that's part of the reason PMF took so much time" (Lenny's Newsletter). Produck's definition follows the HubSpot line: ICP is company-level fit, and it's a filter you apply before you go looking for a persona to sell to.

Why it matters for product-market fit

Product-market fit isn't a single yes-or-no state. It's a moving target that only holds still once you know whose problem you're solving. A vague ICP is the most common reason teams feel like they have PMF with some accounts and are getting ghosted by others. They're actually serving two or three different customer shapes at once, and the signal from each cancels out the others.

This is where an ICP plugs directly into Produck's loop. In the Listen stage, an ICP tells you whose feedback to weight and whose to set aside, so a request from a company that looks nothing like your best accounts doesn't quietly steer the roadmap. In Diagnose, it's the lens for reading a complaint against: is this the customer we're building for, or an edge case. In Decide, it's the tiebreaker when two feature requests compete for the same sprint. In Ship, it's how you check whether the release served the accounts you meant to serve, or just the loudest ones.

We wrote about this connection in what PMF actually means and why it's so hard to pin down. Most founders chase PMF as a vibe, when it's really a fit between a specific product and a specific buyer. Get the ICP wrong and every downstream signal gets harder to read, including feedback from early adopters and your own user interviews. Get it right and the same activities start pointing in one direction.

When it works, and when it doesn't

It works when:

1. You can name the ICP in one sentence a stranger could use to disqualify a lead without asking a follow-up question.
2. You revisit it every few months as your product and your best accounts change, instead of writing it once and filing it away.

It falls short when:

1. The ICP is written to match whoever is easiest to close this quarter, rather than whoever renews and expands.
2. It's so broad it includes almost any company with a website and a budget, which means it isn't filtering anything.
3. Nobody on the team can recite it from memory, so it lives in a slide deck and never touches a real prioritization decision.
4. It gets set once at launch and never gets revisited after the product or the market shifts.

How to apply it

1. Pull your last 10 to 20 customers and sort them by retention and expansion, not by revenue at signing.
2. List the firmographic traits the top quartile shares: industry, headcount, funding stage, tech stack, the problem that brought them in.
3. Write the ICP as a single sentence a salesperson or support rep could use to sort an inbound lead in under 10 seconds.
4. Tag every piece of incoming feedback in Produck's Listen stage against that ICP so you can see, at a glance, whether a request is coming from the customer you're building for.
5. Use the tag as an input in Diagnose and Decide, not the only input, so ICP fit informs prioritization without silencing everything else.
6. Re-run the exercise the next time your retention cohort shifts noticeably, since the ICP that got you to your first ten customers is rarely the one that gets you to your first hundred.