ICP From Closed-Won Data
Derives the ideal customer profile from who actually bought and stayed, rather than from who the team wishes bought.
Free — no card needed
Create a free account
to open ICP From Closed-Won Data — and the other 364 prompts across 21 categories.
Already have an account?
Running it, start to finish
- List retained and churned customers separately.
- Drop any trait that appears in both.
- Convert the survivors into signals you can see before making contact.
What you get back
The output this produces, every time.
- Builds from retained customers and separately from churned ones, which is where the profile earns its keep.
- Drops characteristics appearing in both groups, however intuitive they feel.
- Converts each trait into something observable before a conversation, or reclassifies it as a qualification question.
Getting better results
Where this usually goes wrong, and how to avoid it.
- Look for the trigger event. What was happening when they bought predicts far better than industry or size, and it is usually missing from an ICP.
- Write the anti-profile down. A customer who buys and churns costs more than one who never buys. Knowing who to decline is worth as much as knowing who to chase.
- Split rather than average. If the retained group contains two distinct profiles, one averaged description fits neither and misdirects everything downstream.
More Marketing prompts
All Marketing →- Marketing
Seasonal Campaign Plan
Plans backwards from the buying window rather than from the date, which is usually weeks earlier than people assume.
ClaudeChatGPT001Open → - Marketing
Launch Sequence Planner
Plans a launch backwards from the date across the channels you actually have, with the assets listed and the kill criteria set in advance.
ClaudeChatGPT000Open → - Marketing
Positioning From Customer Language
Builds positioning out of the words customers actually used in reviews, calls and tickets, instead of the words the founding team likes.
ClaudeChatGPTGemini100Open →
Written for The AI University. Every prompt in this library is original work — authored, tested and revised here, not collected from elsewhere. 365 of them, free with an account.