"Niches for riches" with ICPs

How to design a narrow Ideal Customer Profile to build velocity

👋 Hi, I’m James. Thanks for checking out Building Momentum, a newsletter to help you accelerate B2B SaaS growth through go-to-market strategy, sales, and marketing.

💡 Want access to free content, simple tips, frameworks, and deep market insights? I’ll send you two emails every week - just enter your email here:

A misguided founder is asked who the ideal customer is. “Everyone!" That’s the genius of it! Everyone in the world will need this product!”

The same applies to eager early-stage B2B startup marketers, founders, and sales leads. “We will sell to every company in the industry.” Revisit a few months later… and you’ll find are still in the same place. Trying to sell to everyone - SMBs, midmarket, enterprise - without much success.

But those who narrowed their focus on a particular, specific type of customer will usually be further along. They’ll have evidence of traction, be starting to build their ‘mini-brand’, and a pipeline to match.

Why? Niching down brings a multitude of value to early startups: improving alignment, reducing ambiguity across all teams, and bringing people from product to success together to build for the right person. The priority is velocity.

Many startups are reticent to get niche and narrow. “We’ll lose too much time,” or “we need to sell to anyone who’ll take it,” overpowers target-focused teams. And in six months, they wonder why they aren’t seeing any patterns in their data, let alone closed won deals.

Enter the Ideal Customer Profile.

What’s an Ideal Customer Profile?

An ICP is a hypothetical representation of a best-fit customer for your business today. Let’s break it down:

  • Hypothetical - a method for testing and iteration, not a hard rule

  • Representation - a depiction, caricature that’s easily recognisable

  • Best-fit - objective assessment of the type of customers that you can market to, sell to, and support effectively

  • Today - that you can market/sell/support today (or in the very short-term)

ICPs aren’t personas - they don’t describe a person. They describe the company, the firmographics, and any identifying information that makes it easier to recognize them in the marketplace.

An Ideal Customer Profile probably contains information like:

  • Industry

  • Number of employees

  • Revenue

  • Growth rate

  • Geographies

  • Target market (e.g. sells to B2B or B2C)

  • Attributes (e.g. tech stack)

  • Behavior (e.g. company goals and daily activities)

  • Business challenges

ICP fails

I see two common issues with startups using ICPs in their go-to-market strategy.

Balancing breadth

Firstly, the profile has to be broad enough to support marketing and sales activity, yet specific enough to provide a yes/no qualification with a high likelihood of converting to a customer. Unfortunately, the symptom for ICPs that are both too broad or too specific is the same: lack of traction.

ICP as fantasy

Secondly, I’ve seen some teams use ICP as a theoretical device; a dream of the perfect customer… if only they actually existed. An ICP must be based in evidence, or at least be a structured, hypothesis-based attempt to discover a customer segment that will react strongly to your current proposition.

In these scenarios, ICPs become meaningless busywork rather than a strong guiding light for teams across the business.

Bringing structure to your ICP journey

Most ICP advice on the internet just tells you to research your current customers, look for common patterns, and then document it. Easy.

Not so simple… especially if you’re just starting out.

Notion (the VC, not app) has a great resource guide on ICPs, and I’ve used their ICP template multiple times to bring structure to this part of GTM.

Without repeating their post, I’ll summarise the key steps with my tips below:

  1. List the possibilities: get “uncomfortably narrow” and list out all of the segments that you could serve. Use your existing customers, current prospects, market research, and more for inspiration. Segment via industry, revenue, number of employees, geographies, and anything else that can used to create a representation of an ideal customer.

  2. Size the opportunities: add a bottoms-up TAM calculation or a proxy for the size of the market in the Addressable Value column. Tip: use lead-gen tools to search for lists of companies that match the descriptions you’ve created.

  3. Assess Product Market Fit: agree the criteria that match the capabilities of your product with the needs of the customer. Try to keep them simple with Yes or No answers, like “accepts card payments” or “receives 1000s of customer emails each day”. Use an obvious ratings system1 for scores. Average out the criteria score for each segment.

  4. Assess their propensity to buy: using the same approach, list criteria that defines how easy it is to do business with those segments. These criteria might range from simple (are their employees easy to find on LinkedIn?), to complex (does this segment require custom setup? are they willing to do business with a startup?). Use the same scoring system and average it up.

Then, the fun bit: combine the PMF criteria and propensity to buy score for each segment, and explore in conjunction with the addressable value you detailed.

  • Are certain segments bigger markets, but harder to sell to?

  • Are there high PMF segments you could reach today, but don’t have big markets?

  • Do any segments stand out particularly high or low in either category that challenges current thinking?

You’re then left with a prioritized list of Ideal Customer Profiles that you can discuss, make decisions on, and - most importantly - use as a hypothesis to experiment with your go-to-market.

Validating the ICP

Ultimately, your ICP is a bet that you’re making. Are we able to build/market/sell/support these companies effectively? Hopefully, your ICP is based on current activity, and becomes a refinement rather than a redo. Then, validation is less of an issue: continue to do the work, see the results.

For those where an ICP is new and untested, think about the odds of success along the bet. Based on what you know and have done, is the biggest unknown whether you can market to them, or sell to them? This helps understand where to focus your validation efforts. It might be useful to assign a small cross-functional ‘tiger team’ to focus just on ICP customers before rolling out to the rest of the business.

ICP is strategy

An ideal customer profile is a tool to speed up your go-to-market by aligning product, marketing, sales, and success on the type of organization you are building for.

Don’t let it become a stale artefact. ICPs should adapt every six months as you learn more, better understand your performance with segments, and narrow down the ideal customers for your business.

If an ICP doesn’t work out… that’s fine, it’s expected. The ICP is a device to test our hypothesis, whether we can build for/market/sell/support this customer effectively. Look for qualitative signals (lack of excitement and urgency) just as much as quantitative ones (responses to marketing/sales activity, revenue).

Use the ICP as an alignment tool. Embed the profile across the business in conjunction with personas to reiterate the customers that you’re building and solving for (and importantly, who you are not solving for). Velocity - how much is achieved in a certain time period - will increase as a result of reduced ambiguity.

Eventually your ICP will broaden, as a halo effect of success organically introduces you into adjacent segments. But that’s another challenge.

In the meantime, remember: niches for riches2.

Thanks for reading! Let me know what you thought - find me on Twitter @jdomanpipe.

If you enjoyed this post, will you share Building Momentum with your network?

Share Building Momentum


On obvious rating systems; I use 5 for yes or high, 3 for maybe or medium, 1 for low, 0 for no. This create clearer delineations between scores, for easier prioritization.


For non-Americans: ‘niches’ rhymes with ‘riches’ in American English (yep, like “nitches"…)