Ideal Customer Profile
Upload your company list. Get an AI-scored fit report: Tier 1 accounts ranked, feature contributions shown, download-ready.
Up to 700 records · All processing in your browser · No data transmitted
How it works — what the AI scoring actually does
What it does
Scores each company using a weighted linear model across five dimensions: industry fit, company size, revenue, geography, and tech stack. Weights are set by you; the defaults reflect typical B2B GTM priorities.
How attribution works
Each score is decomposed into per-feature contributions using Shapley values, a method from cooperative game theory. The mean feature value across your dataset sets the baseline, and each company's deviation from that baseline is attributed to its features. Contributions sum exactly to the final score.
What it's not
No ML model is trained on your data. The weights are not learned from historical wins; they're your priors. It won't surface non-obvious patterns in your pipeline. For that, you'd need to upload closed-won history and train a model. That's what we build at Agent22.
Read: The Math and Art Behind Your ICP →Sample report 25 companies · SaaS/Fintech ICP criteria
Upload your own CSV to analyze up to 700 real accounts
How to read this report
Tiers
Tier 1 ICP (80–100): strong match across all dimensions. Prioritize for immediate outreach.
Tier 2 Strong Fit (60–79): matches most criteria. Good for nurture with adjusted messaging.
Nurture (40–59): partial fit. Low-touch only until criteria change.
Poor Fit (<40): outside ICP. Remove from active pipeline to improve signal quality.
AI Attribution bars
Each company row shows five mini bars, one per scoring dimension (Industry, Size, Revenue, Geography, Tech).
Cyan bar: positive contribution. This dimension pushed the score up.
Orange bar: negative contribution. This dimension pulled the score down.
Flat bar: neutral (data missing or no criteria set).
Score formula
Below each bar chart you'll see the score broken down:
Base 62 +18 Ind +11 Sz −4 Rev +8 Geo +0 Tech = 95
Base is the average score across your whole list. Each feature's number shows how much it moved this company above or below that average.
Other signals
Three or more dimensions scoring positively with a score above 70. Multiple signals aligning typically means shorter sales cycles.
Feature Importance shows which dimension most differentiated fit across your whole list, not just one company. A tall industry bar means industry was the biggest dividing line in your dataset.
Upload your company list
CSV file · Up to 700 records · Any column names (you'll map them next)
Your data never leaves your browser
All analysis runs in JavaScript on your machine. No CSV rows are sent to any server. No data is stored, logged, or transmitted. End-to-end private by design — not by policy.
Drop your CSV here or click to browse
company_name, industry, employee_count, revenue, country, tech_stack — any column names accepted
500+ records detected. Processing may be slower on older devices.
This tool supports up to 700 records.
Please trim your file to 700 rows and re-upload.
Map columns & define your ICP
Tell us which CSV columns map to each attribute, then set your ideal customer criteria.
Select which CSV column maps to each ICP attribute.
Set your target criteria. Leave blank to skip that dimension.
Comma-separated. Partial match.
Comma-separated keywords found in tech stack field.
Advanced: Adjust scoring weights (defaults: Industry 40%, Size 15%, Revenue 15%, Geo 15%, Tech 15%)
Weights must sum to 100. SHAP values scale with these.
Your report is ready.
Enter your details to unlock your ICP analysis. We'll send you a copy to reference later.
We review every submission personally.
Reply to the email if you want to talk.