I'm assuming you're an indie developer, doing your own (or getting free/partner-based) artwork. I'm also assuming you have no external pressure driving you forward into new games; you can work on your games as much (or as little) as you want, then move on.
Having said all that, how do you decide how much polish to put in a game?
I'm also assuming you won't fall into the notorious trap of dumping your game half (or less) finished; that you have the discpline to work on it until it's "good (enough)."
I currently have a rough cycle where I develop the core functionality, then content/levels/whatever, then add layers of polish (art, bug-fixes, and usability tweaks) until I get bored. There's always scope to improve, but I don't have a measuring stick telling me that "at this point, your return on polishing is not worth the time/effort you're putting into it."
I'm going to assume this is always worth it if you learn something new (like how to render a glow effect, say).
Also, before you ask, I don't have much of a player base (and not one I can observe, either) to ask them when it's good enough.