A reflection on the half-life of side projects, and why the ones I still use share three properties.
Most of the tools I build die within a month. I open them once after launch, share them on Twitter, and never touch them again. The ones that survive — the three or four I still use years later — share a strange set of properties that I only noticed after the fact.
The first is that they solve a problem I have every day. Not every week. Every day. A weekly tool gets re-discovered. A daily tool gets internalised.
The second is that they are smaller than they should be. The temptation, after the first version works, is to add the obvious second feature. Resist. The thing you ship in week one is the thing you will actually use; everything you add in week three is overhead for the parts that already work.
The third is that they have a UI that fits the hand. A keyboard shortcut, a single text field, a one-line command. Anything that requires more than one decision to use will lose to the thing that requires zero.
If you sketch a tool today and you can't honestly answer "do I have this problem every day, and will I open it without thinking" — you are probably building someone else's tool, and that is fine, but call it that.