Posted a 7-slide carousel over the weekend. Decent reach for an account with almost no followers 88% of it from strangers. Then I found the number that actually mattered: average slides viewed was 2.2 out of 7. I'd spent a full day building slides 5, 6 and 7. Almost nobody saw them. Views told me the hook worked. Depth told me everything after it didn't. I'd have made exactly the same mistake again if I'd only looked at the first number. What's the metric you nearly missed that changed how you build?
When user feedback stays in an inbox, it preserves the conversation but makes comparisons difficult. I have found a small research note works better when it records the situation, the observed consequence, and the open decision. It leaves less colour than a full thread, but makes patterns easier to revisit.
What lightweight method has helped you turn scattered user feedback into a decision you can confidently revisit months later?
Half a feature in, agent taps out on usage, and now you're staring at code you didn't fully read yourself. Do you wait for the reset, switch to a different model, or just finish it by hand? Curious how people handle getting cut off mid-flow, not at a clean stopping point.