AutomationApril 11, 2026

I Automated Lead Qualification So My Clients Stop Wasting 4 Hours a Day

Every marketing agency I have talked to has the same problem. Leads come in from five different places and someone has to sit there reading each one. I built an n8n workflow that does all of that in seconds.

Every marketing agency I've talked to has the same problem. Leads come in from five different places — website forms, Facebook ads, landing pages — and someone has to sit there, read each one, Google the company, decide if it's worth replying to, and write a personalized email. For 30-50 leads a day, that's a full-time job.

I built a system that does all of that in seconds. No human involved until there's actually a deal to close.

The Problem

A client of mine runs a digital marketing agency. 40 clients, running paid ads across multiple channels. Their assistant was spending 3-4 hours a day just reading lead submissions and sorting them. Half were junk — no budget, students asking questions, spam.

The good leads? They were sitting in the same pile as the garbage, sometimes for hours before anyone noticed them. By the time someone replied, the lead had already reached out to three other agencies.

What I Built

An n8n workflow that catches every lead the moment it arrives, figures out what it's worth, and routes it to the right place.

1. One door for all leads. Doesn't matter if it comes from Typeform, Facebook Lead Ads, or a website form. They all hit one webhook. A Code node normalizes the data into one clean format — name, email, company, website, message, budget.

2. Company research. The lead says they're from TechStartup BG. The workflow scrapes their website using Firecrawl and pulls back clean text about what the company does. This takes about 2 seconds.

3. AI classification. OpenAI gets the lead info plus the company data and makes a call:

  • HOT — budget over , clear need, mentions timeline
  • WARM — interested but vague on budget or timeline
  • COLD — no budget, generic inquiry, spam

It also writes a personalized email draft. Not a generic template but something like: I saw you're building logistics tools — here's how we can help with your lead pipeline. The AI uses the scraped website data to actually personalize it.

4. Three paths. Hot leads trigger an instant Telegram notification and go into Google Sheets. The team sees it and acts in minutes, not hours. Warm leads get logged for follow-up. Cold leads get logged and ignored.

AI

What It Actually Looks Like in Practice

A lead submits a form at 2am. By 2:01am, the workflow has:

  • Normalized the data
  • Scraped the company website
  • Classified the lead as HOT
  • Written a personalized email draft
  • Sent a Telegram alert to the sales team
  • Logged everything in Google Sheets

The first person who checks their phone in the morning sees HOT LEAD — TechStartup BG — budget 2-5K — email draft ready. They copy the draft, tweak it if needed, hit send. Done.

Compare that to the old way: open email, read submission, Google the company, open a new tab, read their website, go back, think about what to write, type an email, copy to spreadsheet. Multiply by 40 leads. Every day.

The Numbers

  • AI cost: about -15/month for 50 leads per day
  • Firecrawl: free tier covers 500 pages/month
  • Time saved: 3-4 hours per day
  • The assistant that used to do this now focuses on actually closing deals instead of sorting through spam

Tradeoffs

The AI isn't perfect. Sometimes it classifies a WARM lead as HOT because the message sounds urgent but there's no real budget. That's why I don't auto-send emails — the AI drafts them, a human reviews and sends. The AI handles the 80% that's repetitive so humans can focus on the 20% that matters.

Also, web scraping doesn't work on every site. Some company websites are behind Cloudflare or are mostly images. The workflow handles this — if scraping fails, the AI still classifies based on the form data alone. It just can't personalize the email as much.

Try It Yourself

The full workflow is open source. You can import it into any n8n instance and start using it.

GitHub: github.com/vasokai/ai-lead-engine

You'll need API keys for OpenAI and Firecrawl (both have free tiers) and Google Sheets credentials. Setup guide is in the repo.

What's Next

I'm building two more automation projects:

  • AI Document Processor — drop a PDF in Google Drive and AI extracts structured data from it. Invoices, contracts, resumes — it handles all of them.
  • AI Voice Sales Agent — a voice AI that calls leads, qualifies them with a script, and books meetings on Google Calendar.

I'll write about both when they're done.

If you're an agency spending hours on lead sorting and want something like this set up for your specific tools and workflow — reach out.