The feature request that I have been ignoring for a year is the correct one.

There were seventeen individual requests for keyboard shortcuts. I put them all in "nice to have" because there was no frequent use, at most, two times a day.

I shipped them last month in August. It took six hours and included a library and a help overlay. The daily sessions per user went from 2.1 to 3.8 within three weeks.

But it turns out I got causation wrong. People didn't use the app two times a day because it wasn't worth using more than that. They used it two times a day because it took too much effort to open.

It frustrates me that I had the data, saw the ceiling, and misinterpreted it as an indicator of value rather than something else. I would like a better method of doing this.

How do you figure out when you're being disciplined versus when you're being stubborn?