For me, there is a few reasons,
A number of my projects are "because I can", Building something interesting, to learn all the quirks and tricks involved with such a task, e.g. I've never done BGA, someone posted a Kicad project archive asking how to escape
a 12x12 grid one, well I spent 30 minutes and broke it out on 3 layers, respecting grounds,
So now I know how to escape BGA's respect high edge rate return currents, and the fun of 4/4 spacing
Many others get sections ripped out and reused, I've built a few diagnostic / testers over the years, and they are full of reusable, refined and idiotproof chunks,
Others get skipped because I learned enough to discover i was fighting an X / Y problem, or lacked the correct search terms to find a cheap and well made solution.
And a few are simply access to IP preservation, working with some really handy piece of electronics, So I sit down and replicate it in case they later decide to stop selling them, This has become more important recently as a number of small one man band operations have stopped due to old age, (Some of the best RF wizardry I knew of was pumped out by a 97 year old who still hand assembled things up until march when he fell and was whisked away to a home)