You need to figure out what problem your users are having and solve that. Random feature bloat in search of a magical fit is Hope as a Strategy (TM), and it's not typically effective.
Def have to be careful about building features, throwing darts, etc. For this project the genesis was that I wanted to use the product myself and didn't see it in the market; I wanted to scratch my own itch, so I did indeed just build it.
But when I initially did a Show HN[1], the client wasn't far along other than supporting the basic chat feature. By the 2nd Show HN[2], the client had matured a bit and it both withstood the hn blitzkrieg and led to people from all over the world having realtime public conversations - the core idea in action. But people were asking for expected features including tags, search, message permalinks, and other stuff that wasn't there yet.
So that's what I've been building since...
All that said I could be blind to my own biases and it's taken awhile to bring it to maturity. I've also had slowdowns due to outside normal job stuff etc.
What you working on?