Ask a sales team who their ideal customer is and you’ll usually get a sentence. “Mid-market B2B SaaS.” “Fast-growing tech companies in Europe.” Something that looked great on slide four of the board deck.
That’s not an ICP. That’s a vibe.
A vibe can’t tell a rep whether the company on your pricing page right now is worth a call. So every rep decides for themselves, and they all decide differently.
An ideal customer profile (the kind of company that buys from you, stays and pays well) only works when it’s a set of fields you can check and score. This post covers how to write one from the deals you’ve already won, how to turn it into tiers, what to do with each tier and what to do when the data is missing.
Why does a vague ICP cost you?
Because when anything might be a fit, reps chase anything.
The loudest lead wins: whoever filled in the form, whoever replied, whoever happened to be on the website on Monday morning.
A four-person agency in another time zone gets the same attention as a 300-person company that looks exactly like your best customer. (The agency was very enthusiastic on the call.)
Then marketing gets blamed for bad leads, sales gets blamed for not following up and nobody can prove either, because nobody wrote down what good looks like.
The fix is to turn every part of the vibe into something you can check:
A simple test: if two reps can look at the same company and disagree on whether it matches, it isn’t a field yet.
How do you write an ICP from won deals?
Start with who already bought, not who you wish would buy. Your dream-customer list is a wish. Your won deals are evidence.
- Pull the deals you’ve won in the last year or two. Ignore the logo you’re proudest of. One famous customer is an anecdote.
- Write down the same few fields for each: industry, company size, country and tech stack. If you can’t find a field for most of them, it’s not a field you can score.
- Look for clusters, not averages. “Average company size: 340” might describe none of your customers. “Most are between 50 and 200” describes a lot of them.
- Then ask who stayed and who paid. A type of company that buys quickly but leaves after six months isn’t your ICP. It’s your churn forecast.
When I did this for Snitcher, I asked our data four questions: who converts best, who pays the most, who churns least and who actually uses the product. The gap between the top and the bottom was brutal:
- A-tier accounts convert about 3x more than D-tier ones.
- They’re worth about 2.7x as much over their lifetime.
- The top two tiers are a third of our accounts, and nearly 60% of customer revenue.
That’s the difference between “we sell to SaaS” and knowing where to spend your week.
What’s the difference between a strong fit and a potential fit?
Not every field is yes or no. A 150-person software company in the UK might be a perfect match. The same company in Spain might be a decent one: you sell there, you just haven’t won much there yet.
So give each field two buckets, and leave everything else out:
| Field | ✅ Strong fit | 🤞 Potential fit | Everything else |
|---|---|---|---|
| Industry | Software, IT services | Marketing agencies, consulting | Doesn’t count |
| Company size | 51–200 | 11–50 and 201–500 | Doesn’t count |
| Country | UK, Netherlands | Germany, Nordics | Doesn’t count |
| Tech stack | CRM plus marketing automation | CRM only | Doesn’t count |
A strong fit counts for a lot. A potential fit counts for something. That way the “almost” companies don’t get thrown out with the ones that were never going to buy.
Keep it short. Four to six fields that you can actually find for most companies beat fifteen that you can’t.
What’s the difference between an A+ and a D?
Add the fields up and you get a tier. The exact maths matters less than everyone agreeing on what each tier looks like:
| Tier | What it looks like | In plain English |
|---|---|---|
| A+ | Strong fit on every field | Your best customer’s twin |
| A | Strong fit on most fields | Very likely a great customer |
| B | A mix of strong and potential | Worth working if they show interest |
| C | Potential fit on a few fields | Fits on paper, just about |
| D | Little or no fit | Probably not for you |
The top of the ladder is small and valuable. That’s the whole argument for tiers: a few of your accounts deserve most of your attention, and right now they’re getting the same attention as everyone else.
What do you do with each tier?
This is the part most ICP projects skip. A tier is only useful if it changes what someone does on Monday.
Then make it stick. At Snitcher the tier even feeds into commission: reps earn up to 2.5 times more for winning top-tier customers. Turns out reps chase what pays them. (Shocking, we know.)
You don’t have to go that far. But if a D-tier deal and an A+ deal look identical in your CRM, your reps will treat them identically.
What about the companies you know nothing about?
Some companies arrive with half their data missing. No industry, no headcount and a website that’s a logo and a contact form.
It’s tempting to give them the benefit of the doubt. Don’t. Score a missing field as zero, not as “probably fine”. In our own data, companies we couldn’t find anything about converted noticeably worse than the ones we could. A blank cell is a verdict.
Zero isn’t the same as excluded, though. If a no-data company spends twenty minutes on your pricing page, a human should take a look. The tier tells you where to start, not where to stop.
So, what’s your ICP?
Not a sentence in a deck. A short list of fields, each with a strong fit and a potential fit, that adds up to a tier, and a tier that tells each person what to do next.
Write it from your won deals, check it against who pays and stays and look at it again every quarter. It’ll be a bit wrong at first. That’s fine. A vibe is wrong forever.
Curious what tier the companies on your website are right now? Try Snitcher, fill in the scorecard and sort your visitors by ICP fit. Your A-list will be shorter than you think.
