Agent de Scoring ICP

Two Modes, One Agent: Live Scoring and ICP Drift Detection

The ICP Scorer Agent runs in two modes against the same data. Scorer mode evaluates every new lead against ICP the moment it enters the CRM and writes a 0–100 score plus a one-line reason back to the lead record. Drift mode runs monthly and compares the patterns of your actual won deals against the ICP your team documented — when your real wins no longer look like your ICP, drift mode tells you before the board meeting.

How It Works

Scorer mode is hooked to lead-creation webhooks. For every new lead it pulls firmographics (company size, industry, tech stack, funding stage), layers on intent signals if available, and scores the fit against the ICP rubric. The score and the top-three rationale factors are written to CRM custom fields; anything above the team's threshold routes to SDR queue automatically.

Drift mode runs monthly. It pulls all closed-won deals from the last 90 days and compares them against the stated ICP on the four dimensions reps and marketing actually disagree about: segment, industry, company size, and ACV band. When 30%+ of won deals fall outside ICP on any dimension, it fires a flag with the specific cohort and a suggested ICP revision.

Required Connections

• Salesforce or HubSpot — reads leads and won-deal records, writes score fields
• Clearbit, ZoomInfo, or another enrichment source — firmographic fit data
• Slack — drift alerts to the revenue-ops channel

What Makes It Different

Most lead-scoring tools score and stop. This agent closes the loop: the same scoring logic that grades your inbound is audited every month against your real wins, so a stale ICP definition gets caught by the system that depends on it.