Signal sourcing
A public source where every lead had just published a model. The timing is the hook: the email lands while the work is fresh.
Back to case studies
How TPN got 500+ positive replies from cold email
True Performance Network (TPN) compresses, optimizes and benchmarks LLM models for developers and research teams: buyers who skip conferences and never show up in a LinkedIn filter. The first campaign in May got 12 replies on 2,919 emails. closed:in rebuilt the list around what those people had just published, replaced the pitch with a free first step, and the second campaign had 80 interested prospects within 24 hours.
7,128Prospects since 16 Sept 2026
762Replies, 10.7%
466Interested, 1 in 15
24 hTo the first 80 interested
01The challenge
A technical product, buyers who sit in no database
TPN serves developers and developer teams: people who compress, fine-tune and benchmark LLM models. As Mikel puts it, they do not go to conferences, and they do not sit in a LinkedIn title filter either. TPN knew what kind of customer it wanted. The open question was how to reach them.
A previous outbound partner had tried, with a small LinkedIn campaign that got some traction but could not scale. When closed:in started in April 2026, TPN had four possible segments on paper and no proof which one would buy. The first campaign leaned on the obvious angle: cheaper than the big proxy providers. It went out to 2,919 prospects in May and came back with 12 replies and 4 interested. We paused it and went back to the drawing board.
02The solution
Same inboxes, same client, a different offer
The relaunch on 16 September started from a signal instead of a title: people who had just published a model without benchmark scores. That is a moment, not a demographic. The offer became a free first step, closed:in runs the first benchmark for you, with a one-line call to action. Four segments, four campaigns, three plain-text emails, on the same 40 inboxes that had carried the May campaign.
The list came from where those people publish, found and qualified by closed:in rather than exported from a database. What Mikel noticed was the digging: instead of asking TPN for a customer profile and buying a list against it, we went into what TPN does and built the list from the source itself.
Four segments
Four campaigns by what the person had published and who they are, so each segment could be read on its own instead of averaged away.
Offer and copy
No pitch on price. The first benchmark run for free, written as one question, in three plain-text steps.
Infrastructure
20 domains and 40 inboxes, warmed since April, 1,000 sends a day. Nothing new to build, everything to re-aim.
I really like your hands-on approach. Digging in, trying to really understand the business and then finding creative ways to reach customers.
03The results
80 interested in 24 hours, 466 in four weeks
The first 24 hours brought 80 interested replies. After 17 days it was 309. On 13 October, four weeks in, Instantly showed 7,128 prospects contacted, 762 replies and 466 interested: one positive answer for every 15 people emailed, against 4 interested on 2,919 in May.
For TPN the replies were more than pipeline. Mikel counts over 500 positive responses, each one a data point: wish lists, feedback on the business model, input on the platform. Real users uploaded models and requested benchmarks, which stress-tested the product and surfaced bugs the team would not have found without the outreach.
466Interested prospects out of 7,128
1 in 15Positive reply per prospect emailed
80Interested in the first 24 hours
500+Positive replies as product feedback
Want results
like these?
Book a free strategy call and discover how we can build a predictable outbound pipeline for your business.