The LinkedIn Outreach Pipeline
Most LinkedIn outreach fails before you hit send. It's not that your offer is bad, but that the necessary research never happened.
The problem
The default playbook is volume: Connect with enough people, send enough messages, and the law of large numbers eventually works in your favor.
The problem is that LinkedIn is a human-first platform, and the people on it have learned to recognize the formula. Generic opener. Vague value prop. Hard ask. Delete.
The people worth talking to — founders, principals, operators running lean professional services firms — are exactly the people who will not respond to a template. They receive enough of them. What they respond to is evidence that you actually looked.
Most people skip the research because it takes too long. An hour per contact is not a sustainable model when you have thirty names in a list. The LinkedIn Outreach Pipeline solves the research problem, which in turn solves the bland messaging problem.
The distinction
"Hi [First Name], I came across your profile..." is not personalization; it's a mail merge.
Real personalization means walking into a conversation knowing what someone's workflow probably looks like, where the friction likely lives, and what you can say that is specific enough to prove you actually thought about them. That takes research. And research, at scale, is something AI is genuinely good at.
While the pipeline writes your outreach for you, it more importantly builds the foundation that makes good outreach possible.
The solution
Three stages. One system. Built on Google Sheets, Apps Script, and the Claude API, connected to your Gmail and calendar.
You import your LinkedIn connections from a CSV export or acceptance emails. The scoring layer evaluates each contact against a rubric: role, company size, industry, tenure, and relationship signals, plus a technical self-sufficiency penalty that filters out anyone who does what you do. Every contact gets a numerical score and a priority tier. You work the top of the list.
Flagged contacts move to a research queue. The system fetches their website and available public content, then runs it through Claude with a structured research prompt. What comes back: a profile summary, inferred workflow, hypothesized pain points with evidence for each inference, solution angles specific to their business model, and a suggested outreach hook. Not generic. Grounded in how that specific person likely operates.
The research feeds a final drafting pass. The system generates a personalized message, under 150 words, in your voice, with a low-pressure close. You review it in the pipeline, adjust the hook if you want, copy it, and send it from LinkedIn messenger.
One pass of your judgment. The rest is already done.
Where the line is
LinkedIn does not allow direct message automation, and for good reason. The platform runs on human credibility. Every message goes out as you, and it should.
What the pipeline handles is everything that happens before you open LinkedIn: the scoring, the research synthesis, the first-draft thinking. By the time you are looking at a prospect's name, you already know who they are, what their operation probably looks like, and what you are going to say.
You decide what gets sent. The system makes sure that decision is an informed one.
What's included
The delivery model
This is not a SaaS subscription. It is a semi-custom build, handed off to you.
The pipeline is configured around your target profile: your ICP criteria, your scoring rubric, your tone of voice, your offer. It runs inside your own Google account, connected to your Gmail and calendar. You own it outright. No recurring platform fees, nothing that stops working if a vendor changes their pricing.
Setup takes a few working sessions. After that, you have a system that runs every time you want to work a list.
