A comprehensive AI system for lead qualification, conversation quality control, and sales analytics built with n8n, Zoho CRM, PostgreSQL, Google Sheets, and LLMs.

Project details

Context:

• Business/industry: B2B and B2C companies that receive inquiries through a website and work with a CRM.
• Current process: Managers manually communicate with leads, qualify them, create CRM records, analyze conversations, and prepare reports for management.

Problem:

• Initial lead qualification takes too much time.
• Managers qualify customers differently.
• There is no automatic control of dialogue quality.
• There is no unified system connecting CRM, conversation history, and AI analytics.
• Management does not receive regular analytics on the sales team’s performance.

Solution

<aside> 🧠

A comprehensive AI platform was developed with three interconnected workflows that automate the full lead lifecycle — from the first customer message to the CEO’s weekly report.

Tools used: n8n, Zoho CRM, PostgreSQL, Google Sheets, LLM.

</aside>

Solution logic:

Workflow 1 — AI Lead Qualification

• The user starts a conversation with an AI chatbot through a website widget (Chat Trigger).
• The AI Agent qualifies the lead using the BANT methodology, asking one question at a time.
• After collecting enough information, the agent:
  • calculates an AI Score,
  • creates an AI Context,
  • calls the Zoho CRM tool.
• A new lead is automatically created in Zoho CRM.
• The full conversation history is stored in PostgreSQL Chat Memory and linked to the lead via Session ID (Skype_ID).

Workflow 2 — AI Conversation Analysis

• On a schedule, the system retrieves all leads from Zoho CRM.
• Only leads created by the chatbot are selected.
• Using the Session ID, the Chat Memory Manager finds the corresponding conversation in PostgreSQL.
• The AI Agent analyzes each conversation individually.
• For each dialogue, it identifies:
  • AI agent mistakes,
  • the client’s main problem,
  • successful communication patterns.
• The results are automatically written to Google Sheets.

Workflow 3 — AI Sales Monitoring

• Every week, the system retrieves all leads from Zoho CRM.
• A JavaScript Code node calculates the key sales funnel metrics.
• The conversation analysis results are additionally loaded from Google Sheets.
• The AI Agent combines quantitative and qualitative data.
• A structured management report is generated with:
  • funnel issues,
  • bottlenecks,
  • manager workload,
  • improvement recommendations.
• A Telegram bot automatically sends the finished report to the CEO.