My churn dropped 40% when I stopped trying to make users happy and trying to make them confused less

I want to share something counterintuitive that took me two years to figure out.

Like most founders, I was obsessed with user happiness. Did they like the product? Were they satisfied? Did they leave a good review? I'd send NPS surveys, add "how are you feeling?" prompts, respond to every bit of feedback with gratitude.

But churn stayed stubbornly high. People weren't unhappy—they just weren't using the product. They'd sign up, poke around, and drift away.

Then I accidentally noticed something. The users who stuck around weren't the ones who said they loved the product. They were the ones who had figured out the one specific thing the product was actually good at. They weren't confused anymore. They had clicked with the core value.

So I stopped optimizing for "happiness" and started optimizing for "time-to-first-aha-moment." I made the onboarding brutally focused on getting users to that one moment as fast as possible. I removed every optional step. I killed the "tour" and just dropped people into the core action.

Churn dropped 40% in three months.

Happiness surveys didn't change. But confusion went down, and that turned out to matter way more.

Has anyone else found that reducing "confusion" or "friction" moved retention more than trying to increase "delight"? And how do you actually measure confusion in a way that's useful?