14Build an AI Lead Workflow in n8n

The workflow you'll build

Part 1 — Run n8n

docker volume create n8n_data

docker run -it --rm --name n8n -p 5678:5678 \
  -v n8n_data:/home/node/.n8n \
  docker.n8n.io/n8nio/n8n

# requires Node.js 20+
npx n8n

Part 2 — Connect a model

Part 3 — Build it node by node

1. Form Trigger

   Create a workflow → add n8n Form Trigger. Title: “Work with me”. Fields: Name (text, required), Email (email, required), What do you need? (textarea, required), Budget (dropdown: Under ₱20k / ₱20–50k / ₱50–100k / ₱100k+ / Not sure). Click Test step and submit a sample inquiry.

2. AI qualification → structured JSON

   Add Basic LLM Chain (connect your chat model). Turn on Require Specific Output Format and attach a Structured Output Parser with this JSON example:

   {
     "service": "website",
     "summary": "Café owner wants a 5-page site with online menu by November.",
     "budget_php": 40000,
     "deadline": "2026-11-15",
     "urgency": "high",
     "lead_score": 85,
     "red_flags": []
   }

   Prompt (switch the field to Expression to use the {{ }} variables):

   You qualify inbound leads for a freelance web designer in the Philippines.
   Services: websites, branding, SEO. Ideal client: small business, budget ₱30k+,
   clear goal, timeline over 2 weeks.
   
   Score 0–100 for fit. Deduct for: no clear need, budget under ₱20k,
   unrealistic deadline (under 1 week), requests outside services.
   Use null for anything not stated. Never invent a budget.
   
   <lead>
   Name: {{ $json.Name }}
   Budget selected: {{ $json.Budget }}
   Message: {{ $json["What do you need?"] }}
   </lead>

3. Log to Google Sheets

   Add Google Sheets → Append Row. Create a sheet with columns: Date, Name, Email, Service, Budget, Score, Urgency, Summary, Status. Map each column from the form and AI output (e.g. {{ $json.output.lead_score }}). Set Status to “New”.

4. Branch on score

   Add an IF node: {{ $json.output.lead_score }} is greater than or equal to 70. True → hot path. False → nurture path (a polite auto-reply with your portfolio link and booking page — still a good idea to review at first).

5. Draft a personal reply

   On the hot path, add another Basic LLM Chain:

   Write a reply email from Jae, a freelance web designer, to this new lead.
   
   Lead: {{ $('n8n Form Trigger').item.json.Name }}
   Their request: {{ $('n8n Form Trigger').item.json["What do you need?"] }}
   Qualification summary: {{ $('Basic LLM Chain').item.json.output.summary }}
   
   Goals: thank them, reflect their specific need in one sentence,
   suggest a 20-minute call and include the booking link
   https://cal.example.com/jae/discovery .
   Warm, professional, under 120 words. No prices. Return only the email body.

6. Ask for approval on Telegram

   Create a bot with @BotFather, add the Telegram credential, then add Telegram → Send and Wait for Response (approval type: Approve/Disapprove). Message: the lead summary, score and draft. The workflow pauses until you tap a button.

7. Send and update the log

   Add an IF on the approval result. Approved → Gmail → Send to the lead's email with the draft, then Google Sheets → Update Row Status = “Replied”. Disapproved → update Status = “Needs manual reply”.

8. Add an error workflow

   Create a second workflow starting with Error Trigger → Telegram message: “Lead workflow failed: {{ $json.execution.error.message }}”. In the main workflow's Settings, set it as the Error Workflow. Now you'll know immediately if something breaks.

9. Test with 10 realistic inquiries, then activate

   Include: a great lead, a vague one, a tiny budget, an urgent one, one in Taglish, and one spam message. Check scores and drafts, tune the prompts, then toggle the workflow to Active and share the production form URL.

Level it up