Most ICP documents still live in the 20th century with an industry, a company size, a job title, maybe even a trigger event. And that’s why the pipeline looks a bit like roulette: a logistics company on Monday, a fintech on Wednesday, a university on Friday, and not a single company that’s a perfect fit.
That’s not necessarily a wrong ICP but the profile just doesn’t really match the customers who are actually showing up because it doesn’t hold the right details. The worst part is that you already hold the data to see where it works and where it doesn’t.
If you’re the founder of an early-stage B2B tech company facing this challenge, here are a couple of wrong ICP signs I’d check first, followed by a fix that keeps most of your work intact.
Warning 1: Your best customers don’t look like your profile
Your first customers may be quite varied, so your ICP document is probably trying to accommodate them all. It shouldn’t. Your ICP should be focused on the best of the lot.
So list the customers who use the product properly, pay without chasing, and also talk about you to someone else. Then put them next to the profile in your document.
If your best customers and your ICP don’t overlap, the document is wrong and you have to fine-tune it.
At Willow we’ve done that exercise over and over again. There used to be all sorts of companies. But it soon became clear that some customers are happier than others. Some are happy to pay and don’t ask questions. Others don’t value it and always ask more. It’s pretty clear which type of customer you want more of. So that’s the sort of thing you need to take into account. You have to check what those happy customers have in common. For Willow, that was B2B service providers like accountants, lawyers, IT firms, and recruiters. And even then, the group looks rather big. That’s why even within that target audience, you may want to split it up into multiple profiles and write different campaigns, blogs and website pages for each.
Warning 2: You can’t name a company your ICP excludes and why
Try it now. Name one type of company you automatically say no to. If you can’t, you haven’t really thought about an ideal customer profile, or you just don’t respect it.
“B2B tech companies” accepts everything from a three-person startup to a listed software vendor. It filters nobody, so every lead may look like a fit but none of them are a guaranteed sale. Far from it.
Jeroen De Flander has a fitting quote: “You cannot be everything to everyone. If you decide to go north, you cannot go south at the same time.”
And here’s another, more practical insight. Geoffrey Moore’s beachhead idea in Crossing the Chasm says: pick a segment big enough to matter and small enough to lead. Four or five customers in one segment will talk to each other. One customer in each of five segments won’t.
Warning 3: Your deals look nothing like each other
Pull up your last ten closed deals. Look at how long each took, who you talked to, what the projected scope was, and which features came up in the demo.
If one deal closed in three weeks over a LinkedIn message and another dragged for six months through procurement, you have two completely different customer types. The latter may pay a bigger cheque, but the question is if you aren’t better off finding three of the former in the same time?
Here are two patterns that deserve a closer look:
-
Fast deals that churn early. A quick yes isn’t a win if the customer leaves in four months. That’s an optimistic sale because you need the cash, not a long-term strategy.
-
Lots of “no decision” outcomes. These tend to come from companies that like the idea but lack the budget or the readiness to buy. They’re interested without being a fit, and honestly, a waste of your time.
Warning 4: Your pitch changes every time you explain it
Listen to yourself on three sales conversations in a row. If your pitch is different each time, there are two potential explanations. You’re either very good at adapting the pitch to the challenges of the person in front of you or you don’t really know what you’re selling and who you’re selling it to, so you keep changing the story.
A sharp ICP gives you a stronger understanding of the challengers and puts you in the first category.
How to fix your ICP without starting over
The fear behind most ICP rewrites is sunk cost: months of sales pitches, website copy and other assets built on a profile that turned out to be off. I’ve got good news for you: the fix is usually a subtraction, so most of that work survives.
Here’s the order I’d follow:
-
Sort your customers into three groups. Best (stays, uses it, refers), fine, and bad fit. Don’t just focus on revenue. Your best-paying customer might actually lose you money if they are too demanding.
-
Find what the best group has in common. Look for two or three traits: growth stage, a tool in their stack, a trigger that made them start looking. The method from defining your ICP when your best customers found you by accident works here, because it asks about the situation they were in, not the features they liked.
-
Write the negative ICP. Add a short list of who you stop chasing: the company types that churned, discounted hard or needed custom work you don’t offer. Having a clearly defined list of companies that you don’t sell to will do more for your focus than you can imagine.
-
Rewrite the profile around the top group. Don’t be afraid to go into detail. Focus more on what they look for and what they struggle with than how they look from the outside.
-
Sharpen your assets. Point outreach and content at the narrower group and watch whether conversations get shorter and replies get more positive.
What changes once the profile is right
If your ICP work went in the direction, you should be talking to more people who already recognise the problem. And sales should be easier. But in the long run, you’ll also have less churn, happier customers and a business that’s slightly easier to run.
So don’t focus too much on the cash right now. Focus on the people you want to be selling to for the next 10 years.