For me, the answer was to make software that I actually genuinely find useful. Once I got a rough prototype to the point where I was able to get value out of t

For me, the answer was to make software that I actually genuinely find useful. Once I got a rough prototype to the point where I was able to get value out of the software as a user, the project seemed to stick. I naturally return to use the software every few days or weeks or months (because it is useful), and when I do, sometimes I have ideas for improving the UI or what's going on under the hood. So sometimes I knuckle down and spend a day or two enhancing it.