How Do You Turn Closed-Won Notes Into an ICP Update?

How do you turn closed-won notes into an ICP update?

Pull the notes from your last twenty to forty won deals, tag each one for the trigger, the problem, the buyer, and the reason they chose you, then look for patterns that repeat. Rewrite your ideal customer profile around those patterns, not around company size alone. Review it with sales before you change targeting.

Most ideal customer profiles are written once, early, from a mix of hope and a few early customers. Then the company grows, sells to different people, and the document stays the same. Targeting, ads, and outbound keep running on a picture of the customer that stopped being true a year ago.

Your closed-won notes are the cheapest fix. They hold what buyers said, in their own words, at the moment they chose you. The work is turning messy notes into a profile the whole go-to-market team can use.

What should a closed-won note contain for this to work?

A useful note captures why the deal started, what problem the buyer named, who was involved, what alternatives they considered, and why they chose you. If your notes only say "signed, great call," you will need a short retro with the rep for each deal before you can use them. Fix the note template going forward.

Notes vary a lot between reps. Some write paragraphs. Some write nothing. Do not let that stop you. Use what you have, and fill gaps with five-minute conversations with the rep who closed each deal. Their memory fades fast, so do this soon after the quarter ends.

Going forward, add a few required fields to the deal record at closed-won in your CRM, whether that is HubSpot or Salesforce: trigger, primary problem, competitor considered, and main reason won. Short picklists plus one free-text field work best. They make the next ICP review much faster.

Which deals should you include in the review?

Include won deals from the last two or three quarters, and add a handful of lost deals and churned customers for contrast. Exclude unusual one-off deals, like a favor for a friend or a heavily discounted pilot. You want the deals that represent the business you want more of, not every signature.

Twenty deals is enough to see patterns. Forty is better if you have them. Fewer than ten, and you are looking at anecdotes. In that case, run the exercise anyway, but treat the output as a hypothesis to test, not a new rule.

The contrast set matters. A pattern that shows up in wins and losses equally is not a signal. If I see that most wins came through a referral but so did most losses, referral is not what made them win. I covered how to run that comparison with few deals in how to run a win-loss review with only five deals.

How do you tag the notes so patterns appear?

Read each note and tag it on four dimensions: the trigger that started the search, the problem in the buyer's words, the role of the main buyer, and the deciding reason. Use a spreadsheet with one row per deal. Keep tags short and consistent, and merge near-duplicates as you go so counts mean something.

Write the problem tag in the buyer's language, not yours. If three buyers said "our reps waste the morning building lists," tag it that way, not as "prospecting efficiency." Buyer language becomes your messaging later, so preserve it.

You can use an AI model to do a first pass on tagging, especially if you have call transcripts. It helps with volume. But read every tag it suggests before you trust the count. A model will happily invent a neat category that no buyer actually described. I wrote about using AI on sales calls in how to turn sales calls into CRM notes with AI.

What patterns should change your ICP?

Look for patterns in situation, not just in firmographics. A shared trigger, a shared problem, or a shared buyer role across many wins is more useful than "50 to 200 employees." Situational patterns tell you when to reach out and what to say. Firmographics only tell you where to look.

For example, you might find that most wins followed a new sales leader joining, or a move into a new market, or a failed attempt with a competitor. Each of those is a trigger you can watch for. That turns your ICP from a static list of company traits into a set of moments.

Firmographics still matter as a filter. If every win sits in a certain size band or industry, keep that. But layer the situational pattern on top. I wrote about this layer in stop guessing your ICP and look at second-order signals.

How do you write the updated ICP?

Write it as a one-page document with four parts: who the company is, what situation they are in, who buys and who uses, and why they choose you. Add the disqualifiers too. Use the buyer phrases you collected. Keep it short enough that a new rep can read it before their first call.

Each claim in the ICP should trace back to the tagged deals. If you write "best fit is teams replacing spreadsheets," you should be able to point to the rows that show it. That habit stops the document drifting back into wishful thinking.

Include a section called "Not a fit." It is often more useful than the fit section. If most churned customers shared a trait, write it down. Sales will thank you for permission to walk away from deals that look good and end badly.

How do you get sales to accept the new ICP?

Involve sales in the review instead of handing them the result. Ask two reps to tag a few deals with you, then share the patterns before you write the final version. When reps see their own deals in the data, they trust the output. When they get a finished document, they argue with it.

Expect disagreement on edge cases, and welcome it. A rep who says "that pattern is wrong, here is why" is giving you better data. Adjust the profile or mark the pattern as uncertain.

Then connect the ICP to tools. Update the scoring rules, enrichment filters, and list criteria in your CRM so the new profile changes who gets contacted. An ICP that lives only in a document does not change behavior.

How often should you repeat this?

Repeat it every quarter if you close more than a handful of deals a month, or every six months if deal volume is low. Also repeat it after any big change, such as a new product line, a pricing change, or a move into a new market. Each review should take a few hours, not weeks.

Keep the tagged spreadsheet and add to it each time. Over a year, you build a dataset of why people buy from you, in their own words. That is more valuable than most market research you could pay for.

Track what changed between versions. A short change log at the bottom of the ICP shows how your market is shifting, and it helps new team members understand why the profile looks the way it does.

What should you do next?

Export your won deals from the last two quarters, set up a sheet with the four tag columns, and tag the first ten this week. Add the required closed-won fields to your CRM so future notes are usable. Then book an hour with two reps to review the patterns together.

Do not wait for perfect data. Ten honestly tagged deals will teach you more about your customer than another brainstorm about personas.

I build websites and the automations behind them for lead generation, and an accurate ICP is the input every GTM system depends on. If you want help turning your deal history into a profile your team can act on, reach out and let's chat.