This article is part of a resource pack. Want the associated resources, including a PDF guide, sent directly to your inbox? Click here.

Start Phase Building Blocks: ICPs

This article is part of a resource pack. Want the associated resources, including a PDF guide, sent directly to your inbox? Click here.

Disclaimer: content produced from a webinar ran on September 2023 with Harrison covering Ideal Customer Profiles & Personas. Please click the button below to receive a copy of the webinar, slides (PDF) and article.


A critical building block in the “start” phase of building a software business is defining your ideal customer profile (or ICP). Despite a plethora of articles on the importance of understanding your customer & obsessively solving for their needs. Not getting a tight enough definition on who your company should be servicing, as well as those who you shouldn’t be serving… is the most common mistake we see.

In this article we’ll be looking to help you avoid that mistake; looking to define what an ICP is, and how it’s used. Then how you can establish a tight ICP definition for your organization. As well as some examples of both the good and bad! 

We’ll discuss ideal customer profiles in the context of the company. For most b2b organizations we encourage you to start by thinking about the nature and profile of the business that needs your product to begin with. After all, it’s the businesses themselves which are the true “buyer” of your product. Having established which companies are a better or worse fit for your product, we’ll then zero in on the personas that exist within that organization. 

What is an ideal customer profile?

What do we mean when we say “Ideal Customer Profile”? Or ICP? An ICP defines the perfect customer for your organization and/or product.

It’s typically shared, documented and codified as a list of the attributes and qualities that make a company the best fit for the solution you provide. 

What do we use an ICP for? 

Ideally an ICP should help align your org around the type of company you’re trying to build for, message to, sell to. It should inform decisions around all areas of the business. Consider it a guiding light in aligning your product/service/teams around a common customer.

Throughout the running of a business we regularly encounter questions & challenges such as: 

“Should we be servicing this customer?”

“Should we be prioritizing the build of this feature?”

Done well, a tight ICP definition should help your team navigate those tough questions, ensuring you’re building for and servicing your ideal customers.

Typically, the definitions and how ICPs should be used are widely understood. So what are the challenges folks have when trying to define one of their own? 

What are some common mistakes when rolling out ICPs?

No company is “ideal”

Firstly, no company is ever “ideal”. The Ideal Customer Profile is a description or set of attributes describing the perfect potential customer, although never will a company meet every attribute you decide upon

Instead, you should consider customer fit as a spectrum. On one end of the spectrum you have those companies who are a bad-fit for your company, on the opposite end you have those that are a good fit. 

Consider the good & the bad

With customer fit being a spectrum, I’d encourage your team to spend as much time listing the attributes which are true about ideal customers. But also the attributes which are major red flags for the companies you aren’t suited to. 

Ask your teams, “what is true about the folks we don’t want?” Often these are the opposites of your ideal attributes: but some others can emerge.

Think about the end-to-end customer lifecycle & experience

When looking to build out your set of attributes, don’t only consider who is the best fit for the product as it stands today. I.e who can utilise your technology. But also, what is the profile of customer that yields the best sales velocity? Best ACV? Best retention? The customers that results in the least support & headaches?

Throughout this article I've referenced the companies we want to “serve” as opposed to “build for”, or “sell to” exclusively. You should consider the overall experience of the potential customer, the end-to-end lifecycle, hence “serve” is the right word.   

Ultimately, all customers will fit somewhere on the spectrum between bad & good fit. In understanding where any one potential customer falls, you’re able to make the best decision on how or whether to support or prioritise them accordingly.

You will attract non-ICP customers

It’s a common oversight that even if you define a tight ICP, and do absolutely everything you can to only attract only that type of customer. You will inevitably end up attracting some customers that aren’t ideal — particularly if you have a product-led or self-serve business model. 

Hard questions will emerge as to what extent these are supported, built for, etc. Particularly if any of the following are true: 

  • These non-ideal customers may be great in number but not revenue; or
  • They are noisy and demanding (eg. on social, or with feature requests etc).

These decisions can be difficult.

At Paddle - a large portion of our revenue came from a concentrated group of customers. Meanwhile there was a significant long-tail of companies. These folks delivered lower revenues, had high support costs, were noisy, and had very different requirements to those we were proactively targeting in our go-to-market efforts. 

As such, big decisions and debates regularly arose around how we prioritise work for this group. Even whether we support them in the first place.

With these common mistakes laid bare, the next question to answer is how do we each establish our ICP in the first place? 

What is your ICP? Avoid the mistake. 

At the point you’re looking to codify your ICP, you typically have a handful of customers acquired through various means, including founder-led sales.

As the founder, you know in your gut what a good customer looks like vs bad. Over time you just start to get a feel for this — while you acknowledge this understanding will continue to evolve over time.

However - we need to get really really tight at turning this intrinsic understanding of a good vs bad customer, into a clear set of attributes for others to follow.

In order to find more of these great customers, we need to deeply understand them, and clearly articulate this understanding to our team.

One of the reasons I see teams struggling just before or just after Series A, is that a company may have delivered the growth or revenue to justify their latest fundraise. However, they look to hire for and scale GTM without a true understanding of their ICP. 

Going-to-market without this tight definition causes no end of trouble. Wasted dollars, inefficient growth, a customer base too broad to be able to support with retention issues down the line. I’m sure some of this you’ll have read about before. 

A key takeaway from this article is: do not scale your GTM efforts without a clear, and tight understanding and articulation of a good vs bad customer. 

Let’s dive into some examples of the good vs bad. 

How “What is your ICP?” goes wrong.

As founders or leaders we’re asked this question all the time: whats your ideal customer? A common answer may be as follows: 

“We’re targeting businesses based in Europe or the US. Basically, English speaking folks. They have 100+ employees, and we work best in Software & FinTech”. 

This is without doubt an answer many of us will have heard before.

…You’ll be told what countries are being sold to.

…Minimum size of organization is used as a proxy as to whether they have budget.

…Lastly, some industries are listed with some existing customers.

In reality, this ICP could be for anyone, there is nothing unique about it. Perhaps an onboarding tool, collaboration tool, productivity tool, or internal docs? You name it. 

Moreover the list of companies that meet this profile is in the tens of thousands.

What happens next? 

Folks take this huge group of potential customers, with a loosely defined “ideal” customer.

Within this huge group of potential customers, and ill defined ICP, they then over-index on experimenting into the group with different messaging/audiences.

When in reality — they’d be better off narrowing in on the customer they were targeting before trying to optimize in this way.

This is the next most common mistake I’m seeing: folks over indexing on trying to optimize on messaging vs narrowing in on the ideal customer that message should be being delivered to. 

At Notion, they encourage their portfolio companies to get “uncomfortably narrow.”

Messaging vs. Targeting or Incremental vs. Dramatic Gains

Most companies would benefit from spending more time working out who they should be selling to, before they optimize for how they will be selling, i.e. the message, the channel, inbound vs outbound, direct vs indirect and so on.

Theoretically, the gain you’ll get in delivering message A vs B to the right customer should be incremental. Your team will have researched and thought through both messages. Both will be linked to the value you can deliver. Both will work to some degree, maybe in different circumstances.

The gain is incremental - it’s an optimisation.

In comparison, delivering message A or B to a customer who is actually a fit for your product, a company who actually needs it, versus delivering the message to a company that doesn’t. Is much more significant, with greater impact. 

I have occasionally used the example of selling a house for dramatic effect when emphasising this point. 

When looking to sell your property, you want to ensure your listing is as wonderful as possible. You want great photos on a bright day, the square footage (or meters?) to be accurate, and with a vivid description for the prospective buyers. 

But what’s more important than optimising your way towards the perfect listing? 

Ensuring the listing is seen by those actually in the market for a new house. 

This sentiment is exactly the same for our business software. We want to be selling our products to those who want and need them. Get this right first, and only then optimise for what we say to them, how we convert them, etc 

So how do we determine a good ICP?

How to determine your ICP?

In order to determine a tight ICP we need to deeply understand our customers.

The “en vogue” phrase you’d hear around a decade ago was the company “closest to the customer wins”. This statement is absolutely true when it comes to your ICP.

The high-level questions we’re looking to answer are as follows:

  1. What is true above those you’ve won / are winning / happiest / best?
  2. What is true about those you’re losing / churning / contracting / unhappy?

We want to understand the attributes that correlate positively and negatively with the above. 

Eventually we will prioritise companies based on positive attributes, and deprioritise based on the negative. (Prioritisation being a topic we’ve covered with Notion in the past.) 

There are two techniques to uncover some of these attributes. 

ICP Research 

Most obviously, you should start manually researching existing customers and look for commonality.

Remember — we’re looking for positive & negative commonality.

Some ideas and hints for things to look at: 

  • What was true about their team at the time they engaged with you?
  • Were they hiring for specific folks?
  • Were they mentioning specific things in job descriptions, eg. problems they’re solving?
  • When looking at their websites — what makes you get excited? Maybe specific keywords, or a way they describe themselves?
  • Perhaps some other technologies in use?

Deeply research these folks and you’ll start to get a better picture of what is true about them.

To accelerate this process, ask your best performing reps what they look out for when prospecting a customer. What are things that get them excited, or make them consider disqualifying the account?

ICP Discovery

Having done your desk research — we want to speak to your customers directly

  • Ask them — “what was happening in your org when you realised you needed a tool like ours?
  • “What were you reading or researching?” 
  • “Who or what were you hiring for?” 
  • “What was happening or changing in your org?”

Deeply understand this and you’ll start to see what’s true for good vs bad customers. Later down the line, it also helps us identify signals we want to spot when targeting customers in future.

ICP Examples

If we run this process well we should be able to start getting narrower on your ICP.

To give you an example, I’m happy to walk you through’s ideal customer profile. 

For those who don’t know, GoodFit provides commercial data to GTM team. We help map your addressable market of right customers. We enrich these customers or accounts with data interesting to you & bespoke to your business. We then use this data to help you identify the customers most likely to buy right now, and deliver them the best message we can.

Positive Attributes

  • Personas
    • We like to target teams who have RevOps leadership, Sales Leadership, Marketing leadership
    • This is a sign of overall GTM sophistication and investment.
    • Implementation & adoption tends to be better with RevOps
    • Sales/Marketing both existing means there’s more folks to see value out of us and use product
    • It’s even better if one of these folks is new to their role, with new targets to meet. 

Remember: not everyone has all of these things. We sell to plenty of teams with just a sales function, or sales &  RevOps and no marketing

  • Team
    • We’re looking for 4+ XDRs as we feed outbound sales teams.
    • It’s 4+ since “mapping the market” becomes more of a need with multiple reps

The latter is something we actually go back and forth a lot on whether we should look at overall sales team size or min XDRs. Ultimately - this customer fit is a spectrum, and I’m unsure behavior would change regardless.

  • Employee range
    • Listed as ideally being between 100-400.
    • In reality, many of our customers have less 100 employees
    • However, the happiest, best LTV, best usage, best sales velocity tend to have over 100 staff.
    • For companies with over 400 staff, the deals are bigger but sales velocity isn’t as good, and they require more hand holding post-sales.
    • Again, we’re even closing some business with 1000+, but sales velocity is poor.
  • GTM
    • We want an outbound motion established, with email a core channel.
    • Sometimes folks come when building outbound for the first time instead.
    • When we work with these folks — they’re more difficult, big changes to data-set while they work this out… less “good” customers.
    • If mainly selling f2f, or over phones,there is less value in GoodFit.some text
      • Hence wishing for email as a core channel
  • ACV 
    • 10k+ p/a USD
    • Lower value deals, usually correlate with companies that don't tend to invest as much in account data
  • Markets
    • Companies exploring new markets pose similar challenges to companies looking to establish outbound for the first time…some textAlso spotted challenges with retention
    • Want to map EMEA for first time, or new markets, distribute accounts and sell..
    • Risks with retention if experiment in EMEA doesnt work, and we’re not powering their core GTM offering.
  • Technographics
    • CRM: Big indication of sales team sophistication (support all CRMs) prefer HubSpot, SFDC
    • Existing stack: typically great proxy for spend, team sophistication, and also shows they know how to buy software.
    • Bought data before:some text
      • Data is rarely perfect, we know we’re the best on the market, super obvious when comparing… helps if you have something to compare it to.

We see all of these as a spectrum. For no customers are they all true. However, it’s really easy for folks to get a picture of the companies we like to sell to, and who we’re building for. 

Negative Attributes

  • When the main buyer of us is a founder/CEO the dales process is very difficult
    • Much less effective than key stakeholder as VP Sales / SVP RevOps
  • Non-traditional geos
    • Only using us for LATAM, Middle-East makes implementation very slow
    • It is also harder to gather great data in these markets
  • Translations
    • When there is a big reliance on localised titles globally, there is a bigger implementation / support effort
  • Technographic
    • Want to track 10000s of technologies?
    • We can do this, again it just increase cost & effort on our side.
  • Contact Data
    • If you want a tool to give to reps just for phone numbers and that’s it…that’s not us.
    • Qualify out if just after phone numbers.
    • Want to help, map market, prioritize, distribute at an account level - contact data secondary. This is the way we believe folks should GTM.

Few of these things alone are hard disqualification reasons — but a combination of them may result in DQ or not prioritizing work for one of these customers, etc. 

Disclaimer: In this overview I haven’t probably covered off pain/need stuff in above example enough, I focused on attributes commonly used across companies. However, customers of course need to demonstrate a need for your product.  In our case we would prioritise companies further who we know struggling to map their market. These elements are typically quite unique to your business but should absolutely be in there.


Having established what a better or worst fit company for your organization looks like. It’s time to go a layer deeper and think about the personas that exist in this organization. 

you also may like to read
No items found.
you also may like to read

Get the latest from Notion Capital. Sign up to our newsletter.