Sales & Outreach prompt
AI Prompt to Define Your Ideal Customer Profile (ICP)
ChatGPTClaudeGemini
The short answer
Interrogates your best and worst customers to build a usable ICP with buying triggers, disqualifiers and where to find more of them. Copy the prompt below, replace the 6 [BRACKETED] variables with your own, and paste it into ChatGPT, Claude or Gemini.
You are a go-to-market strategist who builds Ideal Customer Profiles from real evidence, not wishful thinking.
CONTEXT
- What I sell: [PRODUCT_OR_SERVICE]
- Price point and sales cycle: [PRICE_AND_CYCLE]
- My best current customers (who they are, why they bought, how they use it, how long they stayed): [BEST_CUSTOMERS]
- My worst customers or lost deals (why they churned, complained, or never closed): [WORST_CUSTOMERS]
- Who I currently target: [CURRENT_TARGETING]
- Markets or segments I'm curious about: [SEGMENTS_TO_EXPLORE]
TASK
If I have provided fewer than 3 best customers, first ask me up to 6 targeted questions to fill the gaps. Otherwise, analyse the patterns and build an ICP.
RULES
- Ground every attribute in the customer evidence I provided. Mark anything inferred as "hypothesis — validate".
- Do not invent market sizes, statistics, or company names.
- Be specific enough to build a list from: "B2B SaaS, 20–100 employees, selling to mid-market, sales team of 5+" beats "growing tech companies".
- Separate the company profile (firmographics) from the buyer persona (person).
OUTPUT FORMAT
1. ICP summary in one sentence.
2. Firmographics: industry, size, geography, business model, tech stack or tools used, maturity stage.
3. Buying triggers: events that make them likely to buy now (e.g. funding, new hire, regulation, growth pain), each with how I could detect it.
4. Buyer personas: economic buyer, champion, and likely blocker — goals, frustrations, what they need to hear.
5. Disqualifiers: signs a prospect looks good on paper but will be a bad customer.
6. Scoring rubric: 6–8 criteria with weights adding to 100 for qualifying leads.
7. Where to find them: channels, communities, data sources, and search filters to build a list.
8. Gaps: what I should validate next and how (e.g. 5 customer interviews with specific questions).What to fill in
Everything in [GOLD CAPS] is yours to replace. Leave one blank and the machine will guess — and the machine guesses like a meatbag on a Monday.
[PRODUCT_OR_SERVICE][PRICE_AND_CYCLE][BEST_CUSTOMERS][WORST_CUSTOMERS][CURRENT_TARGETING][SEGMENTS_TO_EXPLORE]
How to use it
- Your worst customers are the most valuable input. Be brutally honest about why they went wrong.
- Turn the scoring rubric into Clay or CRM fields so the ICP lives in your pipeline, not a forgotten doc.
- Re-run it every quarter with fresh wins and losses. Your ICP is a hypothesis wearing a suit.
Example
PRODUCT_OR_SERVICE: compliance training videos
BEST_CUSTOMERS: three regional healthcare staffing agencies; bought after failing an audit; renewed twice
WORST_CUSTOMERS: a 5-person startup that churned after 2 months; a hospital system that stalled in procurement for 9 monthsWorks with
Tested for ChatGPT, Claude, Gemini. Any modern chat model that follows long instructions will cope; the weaker ones will need you to restate the output format if they drift.
Weapons from the Arsenal that pair well with this prompt:
- ChatGPT — OpenAI's general-purpose assistant: writing, research, images, voice and agents in one chat box.
- Claude — Anthropic's assistant: the strongest long-form writer and careful reasoner in the consumer AI pack.
- Clay — Spreadsheet-style lead enrichment that waterfalls through many data providers, plus AI research agents.