Build the customer profile from closed deals, not from imagination

An ICP is assembled from records, not from a slide

In most companies the ideal customer profile was written from memory: mid-market and enterprise, one hundred to a thousand employees, above-average digital maturity. Nothing about it can be disproved, which is why it is useless: almost any company fits. A working profile comes out of an export of closed deals from the last twelve to eighteen months, and answers one question: how did the deals you won differ from the ones you lost?

What to pull from the CRM

Closed deals only. Won and lost. Open deals prove nothing yet and tilt the sample toward optimism.
Company attributes. Industry, revenue, headcount, location, tech stack, acquisition channel.
Deal attributes. Amount, cycle length in days, title of the initiator, number of people on their side, close reason in plain words.
What happened afterwards. A customer who churned in month three is not a win, it is an expensive qualification error.

Then compare groups. Look not for who buys but for where won and lost behave differently: cycle half as long, larger deal size, lower churn after rollout. If the gap on an attribute is under ten points, leave it out — it is noise that will later send you writing to the wrong people.

Prompt:
Here is a table of closed deals, columns: [list them].
1) Split into won and lost.
2) For every attribute show the win rate inside the group and the group size.
3) Drop any group with fewer than 8 deals.
4) Name the 3 attributes where the win-rate gap exceeds 20 points.
Where the data is thin, say so plainly instead of stating a pattern.

Insight. The most durable marker of a good customer is not size but an internal process your product speeds up. With no process in place you sell a habit rather than a tool, and the cycle triples.

Common mistake. Building the profile from won deals alone. It then describes everyone you ever sold anything to and gives you no grounds to turn anybody down.

Pro tip. Score the group that won and then churned within six months separately. It is the most informative segment you have: it shows who you can sell to but cannot deliver value to.

Cheat sheet

The customer profile is the output of an export, not of a workshop.
Won and lost only; open deals stay out.
An attribute earns a place if the gap is visible and the group holds eight deals or more.
A customer lost in the first quarter counts as an error, not a win.

1. Why are open deals excluded from the profiling sample?

2. Which group of deals usually reveals the weakest point of your profile?

3. Below what group size should you refuse to draw a conclusion?

🔒 Answer the question correctly to move on to the next lesson.