Back to Projects

LinkedIn Outreach Automation with n8n

Sales Navigator scraping → enrichment → AI personalization → scheduled connection requests via Phantombuster bridge → reply detection → inbox handoff.

2026 Automation Engineer Automation

n8nLinkedIn Sales NavigatorPhantombusterClaude APIApolloReply Detection

taimoorakhtar.com/projects/linkedin-outreach-automation

Introduction

Cold outreach on LinkedIn at scale is a discipline problem disguised as a technical problem. Most operators do it manually for 2 weeks, burn out, and stop. The win is building infrastructure that makes 200 thoughtful connections per week feel like 30 minutes of work — without the spam-detection account suspensions.

The Challenge

LinkedIn's anti-automation enforcement has gotten serious. Tools that worked in 2022 (browser-based bots, scraping with rotating proxies) now flag accounts within days. Meanwhile, generic AI personalization ('Hi {firstName}, I noticed you work at {company}') is worse than no personalization. The goal: real personalization at scale without tripping LinkedIn's defenses.

The Solution

Built an n8n workflow that scrapes Sales Navigator results into a Google Sheet, enriches with Apollo for verified emails and recent activity signals, feeds each profile to Claude with a per-prospect personalization prompt, and routes outbound through Phantombuster with conservative pacing (20-25 per day per account, randomized intervals). Replies detected via inbox polling and tagged for human takeover.

Technical Deep Dive

Sales Navigator scraping. Phantombuster Sales Nav search export feeds n8n via webhook. Captures name, title, company, location, recent activity snippet. Deduplicated against a master sheet to prevent re-contact.

Apollo enrichment layer. Each profile enriched with verified work email, company size, funding stage, and recent news mentions. Cuts wrong-targets by ~30% before they enter the personalization stage.

Per-prospect Claude personalization. Profile context + recent activity + company news fed to Claude with a custom prompt that produces one specific reference line + a clear ask. No templates, no fill-in-the-blanks — fresh write per prospect.

Conservative pacing layer. Phantombuster scheduled at 20-25 invites/day per LinkedIn account, randomized intervals between 4-18 minutes, business-hours-only in account's timezone. Zero suspensions across 6 months of operation.

Reply detection and handoff. Inbox polled every 30 minutes. New replies tagged in CRM, Slack notification fired to the operator, automated sequence paused for that contact. Human takes the conversation from there.

Key Features

200+ targeted connections per week per account
Per-prospect personalization, not template merge
Apollo enrichment reduces wrong-target waste
Conservative pacing — zero LinkedIn account suspensions
Reply detection with automatic sequence pause and human handoff

Results & Impact

37% connection acceptance rate (industry avg ~20%)
12-15% reply rate on accepted connections
Zero account flags or suspensions in 6 months
Outreach time investment compressed from ~10 hours/week to ~45 minutes

Lessons Learned

"LinkedIn outreach lives or dies on the personalization line. Spend the AI budget on that, not on volume."

"Conservative pacing is non-negotiable in 2026 — LinkedIn's detection has surpassed most automation tooling."

"Reply handoff to a human is the unlock — automation is for getting the conversation started, not finishing it."

Related Work

Have a similar build in mind?

I'm available for engagement on GoHighLevel implementations, A2P 10DLC compliance, AI automation pipelines, and CRM migrations. Most projects start at $300–$1,200 depending on scope.

Book a 30-min call WhatsApp View More Projects