Orchestrate LinkedIn with n8n
Trigify webhook → n8n → Slack → HeyReach. The sequencer stays dumb.
The loop
- 01Stand up test + prod webhooksPoint Trigify or HeyReach at test until the JSON is boring. POST, not GET.
- 02Filter junk before any AI nodeDrop comments, #no-engage, missing first name. Don’t spend a token on noise.
- 03Draft two sentences, then SlackOpener ends in a genuine question. Null/uncertain output never sends.
- 04Push into a HeyReach campaignAdd Lead to Campaign, not a list. Fallbacks on. Multi-seat auto-rotation.
- 05Replies: sentiment → CRM → Slack → sendHard-code IDs on the first live pass. Calendar only if a time was mentioned.
LinkedIn outbound dies when the tools don’t talk. Umer Ishaq, GTM Engineer at HeyReach, treats n8n as the orchestrator: Trigify (or Clay) fires a webhook, n8n filters and drafts, Slack approves, HeyReach sends. Three plays share one topology — post engagers, ICP + job-post/exit signals, positive replies — and the sequencer stays dumb on purpose.
The guardrails are the product. Test URL first, production URL second, dummy LinkedIn account before live IDs. Auto-throttle if ≥3 errors in 10 minutes. Seat warmup 1–2 weeks. Fallback hit rate over 10% means a Clay column is empty, not that LinkedIn is “dead.” If a play cannot survive a Slack timeout, it is not ready to scale.