Jimmy Kim shows how uploading 30 email sends to Claude or ChatGPT reveals patterns in subject lines, timing, and segment performance in 15 minutes instead of hours. He found a campaign he thought was strong actually failed because the audience was wrong, a mistake AI caught but his gut missed.
AI is helping this massively. You can dump your CSV of your last 30 sends in the Claude or ChatGPT, or you could connect an MCP to it and ask it to find patterns, subject lines, send times, sub-segment performance. It can give you so much information.
This used to take like multiple and multiple hours, dude. I feel like sometimes I was banging my head against the wall on a 4-hour session analysis, but now I want to say it's taking the team in the ballpark of like 15 to 20-ish minutes.
That's slow considering what we'll get to over the next like couple years here, but I think with the analysis getting better, you know what? I like that it's not biased. Like you have your favorite campaign that you want to defend, but AI, they don't care. They're looking at the data and they want to tell you what the data says.
It's a real thing. Marketers get too attached to their work where AI has no ego.
Yeah. I mean, perfect example is like, and I fall into this category too, just as much because we're always proud of our work. But I had a campaign that I thought was a winner because the creative was great. It felt great. It read great. The hook felt great. And I dumped that data into Claude and asked it to look for patterns basically. And it turned out the campaign itself actually kind of underperformed because the segment wasn't that great. Nothing to do with the creative.
Interesting. And that wasn't something that you necessarily maybe would have caught yourself, right?
No, man, I would have never caught it. I would have never seen the segment there. I was just too close to it.
