I've built the first ninety days of outbound many times. I've never once gotten to build day 91.
You're hiring someone to build the system. I write the message first, then build what sends it. The machine behind this email is under it.
Worth a reply?
Erik
attachmentThe first proof run looked great and was lying13 of 28, then 15 of 15, answers that survived an independent recheck, before and after the fix
ReplyReply "stop" and you won't hear from me again.
{
"ts": "2026-09-09T18:20:31Z",
"action": "rendered",
"domain": "heyagentic.ai",
"email": "erik@heyagentic.ai",
"campaign": "erik.heyagentic.ai",
"batch": "p11-sequence",
"position": "1/4",
"delay_days": 0
}I sorted 2,331 restaurants by who owns their ordering channel, shipped it, then got told the vendors do more than my labels said.
So I re-measured all 2,331. One of my counts fell by half.
Would you rather I'd left it alone?
Erik
attachmentI shipped a wrong number, then went back and checked469 to 233, restaurants counted as running a loyalty program, of 968 reachable accounts
ReplyReply "stop" and you won't hear from me again.
Then I sent 25 cold emails to pizzerias, using a real company's published pricing and saying who I was in each one. I cut 14 first: three whose vendor changed, eleven reachable only at a gmail.
Zero replies. The send list is the story.
Worth cutting those eleven?
Erik
attachmentThe last mile is everything that changes between the draft and the send14 of 39, cut before sending
ReplyReply "stop" and you won't hear from me again.
Last one. Before any of this I spent three years at Cleverly reading 200 outbound campaigns and built the framework they were graded on.
My own best campaign replied at 6 percent. None of it was human.
Worth twenty minutes?
Erik
where I have done this before
Cleverly3 yearsRead more than 200 outbound campaigns and built the audit framework they were graded on: targeting, copy, deliverability.
Lumosco-ownerHalf of an outbound agency. Client campaigns, sending infrastructure, and the reporting that says whether any of it worked.
AgenticcurrentOutbound data and systems for eight clients, including Chris Nantista, Jessie Mouton, Lance Parthemore and DJ Baker.
attachmentMy best reply rate was six percent and none of it was human0, of the 29 replies were a human being answering
ReplyReply "stop" and you won't hear from me again.