n8n AI Lead Intake and CRM-Style Tracking Workflow
A reusable n8n workflow template demonstrating AI-assisted lead intake, CRM-style Google Sheets record management, and a human-approval checkpoint using fictional sample data.
Disclosure: Independent workflow demonstration using fictional sample data. This project does not represent client results or a live client deployment.
The Workflow Problem
Most small teams handle lead intake manually — copying submissions into spreadsheets, eyeballing priority, and writing follow-up messages from scratch. The result is inconsistent triage, slow response times, and leads that fall through the cracks before anyone decides whether they're worth pursuing.
The pattern gets worse when outreach is unreviewed. Auto-sending follow-ups without a human checkpoint creates compliance risk and off-brand messaging. This workflow addresses both sides: structured intake with AI classification, and an explicit approval gate before any draft advances to send.
Architecture
Two coordinated workflows handle the full lead lifecycle.
WF1 — Lead Intake
Stateless intake
Triggered by a POST webhook. Validates required fields and email format, checks for duplicate submissions via Google Sheets lookup, then runs AI classification using the Information Extractor node — producing temperature (HOT/WARM/COLD), urgency, business type, a 2–3 sentence summary, and a suggested reply tone.
A 21-column record is written to Google Sheets, a follow-up draft is generated via the Basic LLM Chain, and the draft is stored in the record. WF1 then responds immediately to the caller with the lead ID and the WF2 review endpoint. No waiting, no pause.
WF2 — Lead Approval
Always-on approval handler
A permanently active webhook that receives approve or reject decisions from a reviewer. On approval, it updates the Sheets record to Approved, sends an immediate response to the reviewer, then pauses at a Wait node for up to 24 hours awaiting send confirmation.
When the reviewer confirms, the workflow resumes and routes to Mark Ready to Send. On rejection, the workflow routes directly to Mark Rejected without triggering the Wait node. The approve-then-wait pattern separates the decision from the send action, giving the reviewer a final checkpoint.
Import order
WF4 Error Handler → WF3 Booking Update → WF2 Lead Approval (activate before WF1) → WF1 Lead Intake
This public walkthrough demonstrates WF1 lead intake and WF2 human approval. Additional booking-update and error-handling workflow concepts remain outside the validated public demo until separately tested and documented.
Tools Used
- n8n (workflow automation platform)
- OpenAI gpt-4o-mini (AI classification and draft generation)
- Google Sheets API via OAuth2 (CRM-style record storage)
- Webhook nodes (stateless intake + always-on approval handler)
- Information Extractor node (structured AI output)
- Basic LLM Chain node (follow-up draft generation)
- Wait node (async human-approval checkpoint)
- IF node (approval / rejection routing)
- Stop And Error node (duplicate detection, validation failure)
Human Review and Safety
The workflow is designed so that no outreach action happens automatically. Every path through the approval handler requires a deliberate human decision before lead status advances.
- No email is ever sent automatically at any stage
- All follow-up drafts are stored for human review only
- Approval is required before lead status advances to Ready to Send
- All test data is fictional — no real names, companies, or email addresses appear in the system
- Webhook payloads contain only demo-mode data
What Could This Look Like for Your Business?
The two-workflow pattern — stateless intake plus always-on approval handler — is reusable across any lead or inquiry workflow that needs consistent triage and human review before outreach. The intake webhook can be pointed at a Typeform, a website form, or any tool that sends a POST request. The storage layer can be Google Sheets or another scoped data store, depending on the workflow requirements.
For a client deployment, the follow-up draft would be reviewed and sent by a real team member rather than held indefinitely. The AI classification fields — temperature, urgency, business type, summary — give the reviewer enough signal to prioritize and personalize without reading the raw submission every time.
Have a similar workflow challenge?
Let's map out your intake and approval flow.
Request an AI Revenue Leak Audit to review your current lead or inquiry process and determine whether a structured intake and approval workflow is worth scoping.
Request an AI Revenue Leak Audit