The Cold Email That Gets a 37% Positive Reply Rate: Anatomy of a Real Campaign

If you have run cold email before, you already know the math is brutal. A "good" campaign lands a 1 to 3 percent positive reply rate, and most of those replies are someone telling you to stop. So when we tell you one of our campaigns hit a 37 percent positive reply rate, the honest reaction is skepticism. Good. That number is not a copywriting trick, and you cannot get there by swiping someone else's subject line.

This is a breakdown of the ATI campaign: a retail-tech outbound program that sent roughly 78,000 cold emails and produced over 300,000 CAD in pipeline at a 37 percent positive reply rate. We are going to walk through what actually drove that number, in order of how much it mattered. The email copy is at the bottom of that list, not the top.

The reply rate is decided before you write a word

Most people searching for cold email examples that get replies are looking at the wrong artifact. The email is the visible output. The reply rate is decided upstream, in who you send to and when. If you email the right person on the right day, a plain three-sentence message outperforms a beautifully written one sent to a stale Apollo list.

For ATI, that meant building a targeting layer that did two things: confirmed every contact was deliverable, and waited for a reason to reach out. We run waterfall enrichment across multiple data providers through Clay, so a contact only enters a campaign once we have a verified email and a current role. That is why our bounce rate sits between 0.15 and 0.9 percent instead of the double-digit bounce most teams accept from decayed list data. Bounces do not just waste sends, they tank your sender reputation and pull the rest of your campaign into spam.

The trigger: why this message, why now

The ATI campaign was not a blast to a static list. It was signal-based. We watched for events that meant a prospect had a fresh reason to care, then sent within that window. Funding rounds, hiring spikes, tech-stack changes, and job changes are the four signals that move the needle most, because each one implies a budget or a priority shift.

The relevant part is what a trigger does to the opening line. Instead of "I help retail companies do X," the email could open with a specific, observable fact about the prospect's company that they know is true. That single change is the difference between an email that reads as broadcast and one that reads as written for them. You can see how we wire this up on the signals page. The short version: timing is a multiplier on every other thing you do.

The actual structure of the email

The email itself was short and followed a structure we use across campaigns. Here is the anatomy, line by line:

• Trigger-based opener (one sentence). A specific observed fact about their company. No "Hope you are well." No "My name is."
• The bridge (one sentence). Connect that fact to a problem they likely have right now. This is where the AI personalization layer earns its place: Perplexity pulls context, Claude drafts the connective line, a human reviews it.
• The proof or relevance line (one sentence). A concrete result or a reason we are credible to them specifically, not a list of logos.
• A low-friction ask (one sentence). Not "book a 30-minute demo." Something closer to "worth a look?" that costs them nothing to answer.

That is four sentences. The discipline is in what we left out. Every sentence that does not earn a reply is a sentence that lowers the reply rate.

Infrastructure: the replies have to be readable

A 37 percent positive reply rate assumes the email landed in the inbox. On shared sending infrastructure, industry inbox placement hovers around 60 percent, which means four in ten of your best-written emails are never seen. Our average placement is 98.5 percent, and that is an infrastructure outcome, not a copy outcome.

For ATI we sent from dedicated domains and warmed mailboxes, kept volume per mailbox low, and authenticated everything (SPF, DKIM, DMARC) before the first send. If you want to pressure-test your own copy for spam triggers, our free spam words checker is a starting point, and the full setup is documented on the deliverability page. The point is that copy and infrastructure are not separate problems. A great email in the spam folder has a zero percent reply rate.

Why this is reproducible, not luck

The number that matters here is not 37 percent on one campaign. It is that the same method produced 194 interested contacts for GearLocker off a proprietary database, 177 interested wine importers for Chateau Constellation, and 143 interested physicians for LeverageRx. Different industries, same machine: clean data, real triggers, tight copy, inbox-grade infrastructure.

That machine is what we hand over at the end of a three-month pilot. You keep the domains, the mailboxes, the Clay tables, and the automation. The campaign that hit 37 percent was not a one-off creative win, it was the output of a system you can own and run after we leave.