Lead Classification is Rule-Based Nuance

I spent years managing classic sales and marketing operations. A huge chunk of SDR (Sales Development Representative) hours was wasted on a simple, repetitive task: reading inbound form submissions, googling company size, funding stages, and geographic coordinates, and deciding whether to schedule a call or send them to a nurture sequence.

When Claude 3.5 Sonnet landed, I realized that we could replace this rule-based step with an autonomous decision agent. Here is the step-by-step breakdown of how we rewired the pipeline.

Step 1: Catching the Webhook

We set up a self-hosted n8n instance. The moment a user submits a form on the site, n8n catches the payload via a secure webhook listener. It parses the name, email, company website, and message.

Step 2: Background Data Enrichment

Before passing the lead to Claude, the n8n flow triggers a Node.js scraping script. The scraper queries public APIs and visits the company website to extract:

  • Employee count indicators
  • Primary industry focus
  • Funding updates
  • Competitor tags

    • Step 3: LLM Intent Scoring

    The scraped payload is fed directly to Claude 3.5 Sonnet with a strict system prompt. We instruct Claude to score the lead from 0 to 100 based on company size fit, clear pain points described in the message, and buyer intent. Claude outputs a structured JSON response:

    {

    "score": 85,

    "category": "high-intent-enterprise",

    "rationale": "Company has 150 employees, uses custom database triggers, and wants a sprint audit immediately."

    }

    Step 4: Intelligent Lead Routing

    Based on the score, n8n routes the lead instantly:

  • **Score 80+**: Pushed to the sales team's Slack channel via a webhook, complete with a pre-written personalized email draft. Response time is under 10 seconds.
  • **Score 50-79**: Saved to CRM as a nurture opportunity, triggering an automated email sequence.
  • **Score <50**: Added to the newsletter list, with zero sales reps notified.

    • The results spoke for themselves. In the first month, customer acquisition cost (CAC) fell 31% because reps spent 100% of their time talking to qualified buyers instead of filtering spreadsheets.