n8n Lead Generation Automation — End-to-End Playbook
How to build an inbound + outbound lead gen system in n8n: enrich, score, route to CRM, and book meetings on autopilot.
Most teams treat lead gen as a chain of manual steps: scrape a list, enrich in a spreadsheet, guess at scoring, paste into a CRM, then hope someone follows up. The operators winning right now run one pipeline that does all of it while they sleep.
What the system does
A production lead engine typically has five layers:
- Sources — webhooks, forms, Apollo/Clay exports, or ad platform events.
- Enrich — firmographics, role, tech stack, and contact validation.
- Score — LLM or rules-based fit against your ICP.
- Route — CRM owner, Slack alert, or sequencer enrollment.
- Book — calendar links or voice agent handoff for hot leads.
In n8n, each layer is a node group you can observe, replay, and tune without redeploying code.
Stack I use on client builds
- n8n for orchestration and error handling
- Apollo / Clay for enrichment
- OpenAI for scoring copy and fit reasoning
- HubSpot (or Attio) as system of record
- Slack for human-in-the-loop on edge cases
The workflow JSON I ship with clients starts from a webhook trigger and fans out enrich + score in parallel before a single routing step — same pattern as the Lead Generation Engine case study on the homepage (12.4K leads/mo, 284 meetings booked).
Implementation tips
- Idempotency keys on every inbound event so retries do not duplicate CRM records.
- Dead-letter queue branch for enrich failures — never silently drop a lead.
- Score thresholds as environment variables so sales can tune without opening the graph.
- Observability: log run IDs to a Supabase table or Airtable for weekly funnel review.
Next steps
If you want the starter graph, grab the free n8n template from the homepage hero. For a system built around your ICP and stack, book a discovery call or tell me what you're building.