ICP scoring, explained for B2B sales teams

Almost every B2B company says it has an Ideal Customer Profile. Push a little and you find a slide from a strategy offsite two years ago that says "mid-market SaaS companies in North America." That is not an ICP. That is a vibe. A real ICP is specific enough that you could hand it to someone who has never met your product and have them sort a list of companies into "yes" and "no" with reasonable agreement.

This post is about building that, and then turning it into a score you can apply to leads at volume.

What an ICP actually is

An ICP describes the companies and people you sell to best. Not everyone you could theoretically sell to. The ones where the deal closes faster, the contract is bigger, the churn is lower, and the customer is happy enough to refer others. It is a description of your best-fit buyer, derived from evidence rather than aspiration.

A useful distinction: the ICP describes the company (the account), and the persona describes the person inside it (the buyer or user). You need both. A perfect-fit company with the wrong contact is not a lead yet, and a perfect contact at a company you can't serve is a dead end.

How to define one that's usable

Start with your own customer data, not your imagination. Look at your best 15 to 30 customers and ask what they have in common. Then look at your worst-fit customers (the ones who churned, haggled, or never activated) and note what set them apart. The gap between those two groups is your ICP.

Focus on attributes you can actually observe in a prospect before you talk to them. A great ICP attribute is one you can find on a company page or a LinkedIn profile. Things like:

• Company size. Headcount bands are usually more reliable than revenue, which is hard to find.
• Industry or category. Be specific. "Software" is too broad; "B2B vertical SaaS" might be right.
• Geography. Where you can sell, support, and bill.
• Tech or operating signals. Sometimes a tool they use, a team they have, or a stage they are at.

For the persona, the attributes are role, seniority, and function. Who signs, who champions, who uses the thing day to day.

Write it down in plain language. "Companies with 50 to 500 employees in B2B software, in Europe or North America, where we sell to a Head of Sales or VP of Revenue." That sentence is worth more than a deck.

How scoring works

A score takes the ICP definition and turns it into a number you can apply to many leads at once, so you can rank and prioritize instead of eyeballing. In practice, scoring combines three things:

1. Firmographic fit. Does the company match the size, industry, and geography in your ICP? This is the foundation.
2. Role fit. Is this person a buyer, an influencer, a user, or a bystander? A VP of Sales scores higher than a sales intern at the same company.
3. Signal strength. What did the person actually do? Someone who wrote a thoughtful comment on a relevant post is a stronger signal than someone who left a passing like.

You weight these and produce a single score, often a band like high / medium / low, or a number out of 100. The exact math matters less than the discipline of applying the same logic to every lead.

This is where engagement-based prospecting and ICP scoring fit together neatly. If you are pulling in everyone who engaged with content in your niche, you will get a flood of people. Scoring against your ICP is what turns that flood into a ranked queue. Folkscope scores each enriched lead against the ICP you define so the people worth a rep's time rise to the top, but the concept applies no matter how you collect leads.

A score is not a verdict. It is a way to spend your attention on the right people first.

Common mistakes

Too broad

The most common failure. The ICP is so wide that almost everyone scores as a fit, which means the score tells you nothing. If 80 percent of your leads are "high fit," your ICP is not doing its job. A good ICP excludes most of the market on purpose. That feels uncomfortable, because excluding people feels like leaving money on the table, but a focused ICP is what lets a small team punch above its weight.

Too rigid

The opposite failure. The ICP is so narrow and the rules so strict that real, winnable deals get scored as "no fit" and ignored. This often happens when teams encode every quirk of their best customer into hard rules. The fix is to treat the score as a prioritization tool, not a gate. A low score should mean "later," not "never," at least while you are still learning.

Scoring on data you can't get

If your ICP depends on annual revenue, funding stage, and the prospect's exact tech stack, you will end up with blank fields and guesses. Build the ICP around attributes you can reliably observe. It is better to score well on three solid attributes than badly on eight aspirational ones.

Setting it and forgetting it

Your best-fit customer changes as your product and pricing change. An ICP from two years ago may be quietly wrong. Revisit it when you notice the score is no longer predicting good deals.

How to tune it

You tune an ICP the same way you built it: with outcomes. After a few months, look at which scored leads actually became opportunities and customers. If your "high fit" leads close at a much better rate than "medium," the score is working. If there is no difference, your weights are off or your attributes are wrong.

A simple, honest tuning loop:

• Tag where each closed deal sat in your scoring when it entered the pipeline.
• Check whether higher scores correlate with better win rates and bigger deals.
• Adjust weights toward the attributes that predict good outcomes, and drop the ones that don't.

You do not need a data science team for this. A monthly look at "did our high-fit leads do better than our low-fit ones" will tell you most of what you need.

What it adds up to

The point of ICP scoring is not precision for its own sake. It is giving a rep a ranked list instead of a pile, so the first hour of the day goes to the best-fit person who showed interest most recently. Define the profile from real customer evidence, score on attributes you can actually see, keep it focused, and adjust it when the outcomes tell you to. Do that and the score becomes something your team trusts instead of a slide nobody opens.