Most ICPs are built in a workshop from guesses and anecdotes. The result is a target audience that matches who you think your best customer is, not who they actually are. The gap is measurable — accounts that match a data-derived ICP close at 2–3× the rate of accounts that don't.
Clawdbot analyzes closed-won deals, identifies the firmographic and behavioral patterns that correlate with revenue and retention, and generates a target list that matches those patterns.
ICP vs. Persona
| ICP | Buyer Persona | |
|---|---|---|
| Unit | Company | Individual |
| Describes | Which companies to target | Who within the company buys |
| Built from | CRM + revenue data | Sales call transcripts + interviews |
| Used for | Prospecting, ad targeting, list building | Messaging, content, objection handling |
| Changes | Quarterly review | Rarely — unless category shifts |
Build the ICP first. The persona is useless if you're talking to the wrong company.
Data Inputs
Clawdbot needs three datasets from your CRM to derive the ICP:
Closed-won accounts (last 12 months)
Company name, industry, size, contract value, sales cycle length
Churned accounts (last 12 months)
Same fields — ICP is defined by what churns as much as what converts
Expansion accounts (highest LTV)
Companies that expanded — these are your best-fit customers, not just your first customers
Firmographic Signals
These are the firmographic variables Clawdbot tests for correlation with high LTV:
Behavioral Signals
Firmographics tell you who. Behavioral signals tell you when:
Job posting for a role your product replaces — active pain signal
New VP or C-suite hire in past 90 days — evaluation window
Recent fundraising round — budget exists, evaluation mode active
Series A/B stage — large enough to have the problem, small enough to move fast
25%+ headcount growth in 12 months — scaling pain is real
Prospect List Building
Once the ICP is defined, Clawdbot builds the prospect list via:
Apollo.io
Filter by firmographic ICP + behavioral signals; export contact list
Clay
Enrich with real-time signals — recent job postings, funding, news mentions
LinkedIn Sales Navigator
Company filters + decision-maker identification by seniority + function
Clawdbot Configuration
| Setting | Value |
|---|---|
| Input | CRM export: closed-won + churned + expansion accounts with firmographic fields |
| Analysis | Correlation scoring: which firmographic variables predict LTV > median |
| Output | ICP definition doc + weighted scoring criteria + prospect filter spec |
| List generation | Apollo/Clay filter spec for immediate prospect list build |
| Scoring model | Each prospect scored 1–10 against ICP criteria |
| Refresh | Quarterly — ICP shifts as your product and market evolve |
Foundation
Haven't set up Clawdbot yet?
OpenClaw + Telegram + Claude. Takes ~20 minutes.
