http://news.ycombinator.com/item?id=83660
It's growing pretty well now, handles around 10,000,000 messages a day. As you can see from the original post, I started out having a telnet app, irc app, and game app (Multiplayer jigsaws!). The biggest decision I made was to focus on one. I chose IRC for obvious reasons. The feedback I got was that people didn't really know what it was etc - hence the decision to focus on one app.
To be honest I was pretty gutted about the feedback, and was about to give up on it. I then decided to post it to reddit anyway (nothing to loose) a little later, and got a few people using it there. I then found ircatwork.com which I bought pretty cheaply giving me some users to start with. I naively thought I suppose, that posting to hacker news and reddit would lead to users. It didn't. Sure, a little spike in users that lasts a day or so, but it's not long term.
The biggest obstacle so far is probably getting to a state where you can make money. Premium accounts, targeted ads etc all take time.
Also in my case, dealing with IRC networks has been an eye opener. Quite a lot of the older networks are completely opposed to anything like Mibbit... "But then anyone could connect!" they say. Most are friendly now though, and at peak times we have 350 or so connections to freenode, and about 2,500 IRC connections open in all.
It's growing well, but that means it's costing a little more to run, and not yet making money. But I'm sure it'll sort itself out :/
I think the main thing is to grow and get user feedback, and that seems to be happening ok.