Any teams who need an easy, step-by-step way to define their Ideal Customer Profile and stop wasting money on the wrong prospects.
No ICP → no clue what to clean. ICP tells you which fields matter (industry, employee count, region, buyer role, etc.), so you can:
If your CRM can export Deals/Opportunities, Accounts/Companies, and Contacts/People to CSV, you can run this. Names vary by platform; the logic doesn’t.
Not an absolute requirement, but you can pick the outcomes your business actually cares about. We’ll measure segments against these:
Pull closed-won and closed-lost deals from the last 12–24 months. Include:
💲From Deals/Opportunities: deal_id, account_id, amount, stage, close_date, created_date, won/lost, loss_reason, product_line, source/campaign, owner, competitor (if tracked).
🏢From Accounts/Companies: account_id, industry, subindustry, employee_count, annual_revenue, country, state/province, zip/postal, hq_region, tech_stack (if you track), customer_tier, plan, ARR (if post-sale data is available).
👤From Contacts/People (associated to the deal): contact_id, account_id, job_title, department, seniority, persona (normalized if you have it), role_in_deal (economic buyer, champion, user).
Join on IDs so you can analyze deal + account + primary buyer in one row. If your CRM can’t export a joined file, export three CSVs and join in Excel/Sheets (VLOOKUP/XLOOKUP on IDs) or any BI tool.
❗Before anyone analyzes the file: remove anything that can identify a specific company or person.
account_id, deal_id, and the non-identifying fields above.🗝️Why? If the spreadsheet walks out the door, your customers don’t. You can always map insights back to real records later via IDs. 🗺️📍
You don’t need perfection; you need “correct enough to rank segments.” Fix the landmines:
Financial Services).C-suite, VP/Director, Manager, IC, Ops, IT, Finance, Sales, Marketing.Unknown, don’t invent.Pro tip: When you fix values here, re-import the corrections back into your CRM using the IDs. Don’t fix the same bad fields every quarter. (Unless you like wasting weekends with Excel and existential dread.)
Pick your AI tool of choice (ChatGPT, Gemini, Claude, etc.). Upload the masked CSV and paste the prompt below.
Notes:
Pro tip: Use your AI’s deep analysis / advanced data analysis / code interpreter mode for this step. Regular chat will skim your CSV and hallucinate confidence. You want the version that actually ingests files, runs calculations, and returns tables.
You are an expert RevOps analyst. I will give you a CSV where each row represents a closed deal (won or lost) joined to account and primary contact attributes.
Your job:
1. Ask up to 5 clarifying questions to align on goals:
- Should we prioritize win rate, ACV, or cycle speed?
- Any strategic exclusions (e.g., churned accounts, internal deals)?
- Focus on specific products, geos, or personas?
- Timeframe focus (e.g., last 12–18 months)?
- Do you need to confirm field mappings?
2. Normalize and validate field names:
- Detect common CRM variants (e.g., created_date, created_at, deal_created; close_date, closed_at, deal_closed)
- Ask for confirmation if any key fields are ambiguous or missing
3. Analyze segments across available categorical dimensions:
- Examples: industry, employee_count, annual_revenue, country, state/province, product_line, source/campaign, job_title, department, seniority, persona, tech_stack, competitor, customer_tier, plan, and any other categorical columns provided
4. For each segment and smart combinations (min n=30), compute:
- win_rate = wins / (wins + losses)
- avg_deal_size (wins only)
- avg_sales_cycle_days = avg(close_date − created_date)
- churn/expansion rates if present
- Flag outlier deals (>3× median for deal size or cycle time) and note their impact
5. Identify:
- Top 3–5 ICP segments (high win rate and/or ACV, acceptable cycle time)
- Anti-ICP segments (low win rate, small deals, long cycles, high churn)
6. Detect data quality issues:
- Missing or ambiguous values (e.g., empty industries, unknown job titles, malformed dates)
- Inconsistent formatting
- List specific records or patterns that need correction
7. Recommend:
- Minimum viable CRM fields to track for ongoing ICP validation
- Picklist values to standardize (e.g., titles, industries, source)
- Predictive actions to improve performance in anti-ICP segments
Output Format:
1) Executive Summary (bullets a VP can read in 60 seconds)
2) Detailed Report (Markdown, print-friendly for PDF export):
- ICP definition table with metrics
- Anti-ICP table
- Segment comparison tables
- Outlier callouts
- Data hygiene findings & fixes
- Operational playbook: routing, scoring, sequences, targeting, required fields
Notes:
- If data is thin (n < 30), avoid over-fitting
- If required fields are missing or mappings unclear, ask for clarification
🔮Power moves to make with the resulting data:
deal_id,account_id,contact_id,won_lost,amount,created_date,close_date,loss_reason,product_line,source,campaign,competitor,industry,subindustry,employee_count,annual_revenue,country,state,postal_code,hq_region,job_title,department,seniority,persona,role_in_deal,tech_stack,customer_tier,plan,arr
If you want this done without the homework, CRM Whispers can run the export, masking, cleanup, AI analysis, and CRM changes for you — using your stack.
▶ Option A: Schedule a 20-minute consult Schedule 20 Minute Chat.
▶ Option B: email us at info@crmwhisperers.com
You tame ICP first. Next tame the data. Then the revenue follows. And if you ignore the warnings above, you also tame an HR investigation . 🙁