LinkedIn Warm Outreach: Book Calls Without Cold DMs

Key takeaways

Your LinkedIn Connections export is free, first-party, and 1st-degree only: Me > Settings & Privacy > Data Privacy > Get a copy of your data. It arrives as a CSV in about 10 minutes to 24 hours. Delete the notes rows above 'First Name' and the columns snap into place.
The export is a shortlist tool, not a message source. It gets you from 1,400 people to 20. Then you close it and read those 20 live profiles, because name plus company plus title is far too thin to write anything a human would answer.
Vendor-reported benchmarks put 1st-degree reply rates around 16.9%, versus roughly 2% to 5% for a genuinely cold DM. The same data says an out-of-contact 1st-degree connection replies more like 5% to 15%, so personalization is what matters.
'Connected On' is the most useful column. Very recent and very old both score well. The dead middle, connected a year ago and never spoken to, is the worst of both worlds.
Run every message through the swap test and the screenshot test before you send. If another name still fits, or you'd wince at it being posted publicly, it isn't ready.
The most valuable output of any ranking pass is the do-not-message list. The names you cross off protect relationships you spent years building.
LinkedIn automation violates the User Agreement (Section 8.2). In March 2026 LinkedIn removed vendor HeyReach's company page and banned executives' personal profiles, while the software kept running for customers. Nobody outside LinkedIn knows why it happened: the cloud-proxy explanation you'll read everywhere comes from competing vendors, not from LinkedIn and not from reporting.

The Solo Creators AI Studio

Get this skill, and every one I build.

The whole vaultA new drop monthlyBuild live with meYours to keep, forever

See inside the Studio

30-day money-back guarantee. Cancel anytime.

Questions people ask

Is exporting my LinkedIn connections against LinkedIn's rules?

No. It's LinkedIn's own built-in feature operating on your own data, available to free accounts. Go to Me > Settings & Privacy > Data Privacy > Get a copy of your data, tick Connections, and request the archive. It arrives by email as a ZIP with Connections.csv, usually within 10 minutes to 24 hours. What violates the rules is scraping profiles you aren't connected to, or automating activity. Downloading a list of people who already accepted your request is neither.

Can I just write my LinkedIn messages straight from the CSV export?

No, and this is the mistake that ruins the whole method. The export gives you a name, a company, a job title, and a date. That's far too thin to write anything a human would answer, and a message built from it reads exactly like what it is: a mail merge. Use the CSV for what it's good at, which is narrowing 1,400 people down to about 20. Then open those 20 profiles and read them. The spreadsheet picks who. The profile tells you what to say.

Why are most of the email addresses in my Connections.csv blank?

Because LinkedIn only includes someone's email if that person has switched on 'Allow connections to export my email,' and it's off by default. So most rows have nothing in the Email Address column. That's the system working as intended. Don't treat the export as an email list or try to fill the gaps with a lookup tool. Use the URL column, go to their profile, and message them on the platform where they actually agreed to hear from you.

My Connections.csv opens as one giant messy column. What did I do wrong?

Nothing. LinkedIn puts a few notes rows above the real header, so spreadsheet apps guess the structure wrong. Delete the rows above the one that starts with 'First Name' and the columns line up immediately. If it still looks wrong, re-import the file and tell your spreadsheet app it's comma-separated. That's the entire technical difficulty of this method, and it's why nobody needs a paid tool to do it.

How many LinkedIn messages can I safely send in a day?

Connection invitations are capped at roughly 100 per week on a rolling 7-day window, with a daily soft cap around 20 to 25 before throttling. Messages to existing 1st-degree connections aren't invite-capped, but sending them in a rapid burst can still trip spam detection. A practical rule is five to ten warm messages a day, sent by hand and spread out. That's well inside every limit, and it's more follow-through than most people sustain anyway.

Is a tool like HeyReach or Dux-Soup safe to use in 2026?

Automated activity, bots, scraping, and browser extensions that act for you are prohibited under LinkedIn's User Agreement (Section 8.2). In March 2026 LinkedIn removed HeyReach's company page and banned its executives' personal profiles. LinkedIn never said why, and the cloud-proxy explanation that circulates comes from competing vendors rather than from LinkedIn or any reporting. Notably it wasn't any user's conduct. Notably the software kept running for customers, so the people who lost their LinkedIn presence were the vendor's own team. That's the real lesson: your compliance isn't the only variable, the vendor's architecture is.

Should I buy Sales Navigator instead?

If you're breaking into a market where you have no 1st-degree connections at all, yes. You can't export relationships you don't have, and Sales Navigator is built precisely for finding qualified strangers at scale. It's LinkedIn's own product, so unlike third-party automation it carries no ban risk. The mistake is buying it while 1,400 people who already accepted your connection request sit unread. Work the warm list first, because it's free and it replies at several times the rate. Buy the subscription when you've genuinely run out of warm.

Does using AI to write the message make it fake?

The message is fake when the detail is fake, not when AI helped you phrase it. If you genuinely connected with someone in 2023, they genuinely posted about changing their business, and you genuinely found it interesting, AI tightening that into four honest sentences is a writing assistant. AI inventing a project they never worked on, or mail-merging one compliment to 200 people, is the fake thing. Always make the model tell you which detail it used and where it found it, then go verify it on the profile before you send.

Found this useful? Share itPost Share

Written by Ronnie Nijmeh

Ronnie spent 18 years building a SaaS with a team of 20 that served over 650,000 customers, generated over $14M in sales, and sent over 550M emails. Now he's solo, solving real business bottlenecks and turning them into working AI skills, workflows, and automations. He teaches all of it, with direct access to him, inside the Solo Creators AI Studio Skool community. See what he's built →