I've shipped a lot of side projects, and most of them died. The ones that survived had a few things in common.

What mattered:

• A single painful problem I personally had
• The ugliest possible UI that still worked
• One deploy command

What didn't matter:

• The perfect stack
• A landing page before the product
• Tests for the first version (controversial, I know)

The last one I'd qualify: no tests on day one, but write one as soon as real users show up.

My current side project is a small tool that turns markdown notes into a static site. TypeScript and Go, one Postgres database, one deploy. It's not clever, but it's mine.