Lead Qualification Automation (n8n)

Automated lead qualification pipeline using AI classification.

What it does

1. Triggers when someone submits a Google Form (demo request)
2. Validates that the message isn't empty
3. Classifies the lead using OpenAI:
   • ACTIVE_BUYER (hot)
   • EVALUATING (warm)
   • TIRE_KICKER (cold)
   • UNCLEAR
4. Routes based on classification:
   • Hot leads → Slack alert
   • All leads → Logged to Google Sheet

Workflow Screenshot

Tech Stack

• n8n (automation platform)
• OpenAI API (GPT-4o-mini)
• Google Forms (input)
• Google Sheets (logging)
• Slack (notifications)

Setup Instructions

1. Import workflow.json into n8n
2. Connect credentials:
   • Google (Forms + Sheets)
   • OpenAI API key
   • Slack
3. Create a Google Form with fields: Name, Email, Company, Company Size, Message
4. Create a Google Sheet with columns: Timestamp, Name, Email, Company, Company Size, Message, Intent, Urgency, Reasoning, Route
5. Activate workflow

AI Prompt Used

See the OpenAI node in the workflow. Key design choices:

• Fixed categories (prevents hallucination)
• JSON-only output (parseable)
• Low temperature (0.2) for consistency

Failure Modes Observed

Issue	How I Handled It
Empty form submissions	IF node skips AI call
Malformed JSON from AI	Code node catches parse errors
Inconsistent classification	Low temperature + explicit categories

What I Learned

• How to chain n8n nodes with conditional logic
• Prompt design for structured output
• Importance of logging for debugging AI workflows
• Building fallbacks for when AI output is unexpected

Improvements I'd Make

• Add email notification for WARM leads
• Connect to a real CRM (HubSpot/Salesforce)
• Add human review queue for UNCLEAR classifications
• Track classification accuracy over time