// LINKEDIN LEAD GENERATION
LinkedIn Post Engagers: 81% Acceptance on 2,499 Requests
Cold lists typically get 10 to 20% of LinkedIn connection requests accepted. We wanted to know what happens when every request goes to someone who just showed interest in the topic. So for four months we only contacted people who had liked or commented on a post in our niche. No bought lists, no title-only filters.
The results
| Stage | Count | Rate |
|---|---|---|
| Connection requests sent | 2,499 | |
| Accepted | 2,026 | 81.1% of requests |
| Replied | 943 | 46.5% of accepted connections |
| Interested | 383 | 40.6% of replies |
| Benchmark: unscored cold lists | 10 to 20% accepted |
Campaign period: May 29 to September 25, 2026. "Interested" counts only clear positive replies, not neutral replies, objections or information requests.
Why post engagers convert
- The topic is on their mind this week. They reacted to it in public.
- They are active on LinkedIn, so they see your request.
- They will recognize what you are talking about, so your note reads as a continuation, not an interruption.
How the pipeline works
- Search posts by topic, not single viral posts: topics keep fresh engagers coming every day.
- Pull everyone who reacted or commented and remove duplicates across posts.
- Drop obvious non-fits with free rules on headline and company before any AI call.
- Score each remaining person 1 to 10 against the ICP with a one-line reason.
- Send a connection note built from the person's role and, for commenters, their own words.
- Continue the conversation in one inbox, toward a free trial.
What we learned
- Volume comes from topics, not from single posts. One viral post runs dry in a day.
- Commenters beat likers. A comment takes effort and gives you the best opening line.
- Free filters before AI keep the cost per qualified lead low.
- High acceptance protects the account: fewer requests per new connection means fewer pending invitations.
With AtomSell this whole pipeline is one prompt in Claude: find the posts, pull the engagers into a table, score, write the notes and send. The same signal that worked in a one-week pilot held up over four months and 2,499 requests.
On your current stack
To run this without AtomSell you would stitch together four or five tools. A scraper actor to find posts by topic, because Clay's post audiences only cover posts about your own brand. Separate actors for reactions and for comments, each on your LinkedIn session cookie and maintained by a different developer. A Clay table or a spreadsheet to merge people across posts and remove duplicates, with credits spent on every row. AI columns to score them. Then an outreach tool such as HeyReach or Expandi to send, with its own inbox. Data moves by hand between every step, and the pipeline stops whenever one scraper breaks.
With AtomSell it is one prompt and one table: find the posts, pull the engagers, score, write and send, and every reply comes back to the same place.
Run it yourself
Start a 3-day free trial with full access at app.atomsell.ai/register, connect AtomSell to Claude, ChatGPT or Cursor, and run the playbook on your own buyers.