High-throughput pooled DNA libraries · LinkedIn outbound
Finding leads by what they post, not their job title
01
Why searching by job title does not work
No job title tells you who is about to build a DNA library. Searching by title and company returns thousands of people, and almost none of them have the problem right now.
The audience is small enough that burning it with generic outreach cannot be undone. There is no second list.
What someone posted about last week does tell you. It is the only signal available that shows what they are working on rather than what they are called.
02
How it works
Fig 1 Finding the posts, then the people who engaged with them
Runs on a schedule into Apify, then the LLM throws out the posts that are not real discussion. Everyone who liked or commented on what is left gets pulled out, duplicates removed, and sent to Clay to be enriched and scored.
Approved leads move from the Google Sheet into HeyReach every 3 days.
An LLM cleans up the names first, stripping titles and credentials, because “Hi Dr. Sarah Chen PhD” instantly reads as automation.
Fig 2 Two different openers, depending on who they are
The person who wrote the post gets a different first message from the people who engaged with it. Both send in small batches with a pause in between.
Fig 3 What happens when something breaks
3 nodes, and the order of the middle 2 is the whole point. It switches the broken workflow off before sending the alert email. The other way round, a broken run keeps messaging people while the email sits unread.
03
Results
42.6%message reply rate
78.5%connection acceptance
20replies from 47 messages
3 daysbetween sending runs
One month of an ongoing campaign: 65 connection requests, 51 accepted, 47 messages, 20 replies. Those replies became booked meetings, which is the number that matters to them.