T
23

Spent 9 months tweaking AI ad copy before realizing the model was the problem, not the prompts

Every week I'd adjust tone, length, and CTA phrasing in Claude and Gemini, but our click-throughs stayed flat until a data guy in our Chicago office ran a quick A/B test and pointed out the baseline model was 2 versions old, has anyone else burned months on prompt polish when the real fix was just swapping the engine?
0 comments

Log in to join the discussion

Log In
0 Comments

No comments yet

Be the first to share your thoughts on this discussion.