I built an AI outbound agent and I run my own outbound with it every day. When I started, I spent most of my time on the emails themselves: subject lines, length, the opening line. That stuff matters. But the thing that changed my results was learning who my real audience was.
Every founder has a picture of their ideal customer. Mine looked great on paper. The replies showed me it was only partly right. Some groups I was sure about stayed quiet. Others I’d barely considered turned out to be exactly the people who needed what I sell.
You can’t see that from a spreadsheet. You only see it from replies.
So now my outbound works like a learning loop:
Start with a few different groups instead of betting on one “perfect” list.
Treat every reply as data: who answers, who asks questions, who says “not now”, who never responds.
Shift more emails toward the groups that respond, week by week.
Then add signals on top: a launch, a new hire, a funding round. Once you know who to write to, signals tell you when, and that’s where it gets really sharp.
After a few weeks, the audience you’re writing to looks different from the one you started with, and it’s the right one.
One founder’s notes, not a law.
Did your real audience turn out different from the one you first had in mind? How did you find out?
