I'm primarily a designer and composer so I get simple gameplay mechanics up but generally work on establishing a mood first. I've found that once I do this, themes, gameplay mechanics etc will start to fall into place. It doesn't happen all at once, it's an ongoing process of constant refinement and doesn't need to be rushed, it will happen naturally if you continue to immerse yourself in the project.
Some people like to plan things out meticulously first, I don't. There's no right or wrong way, it's what works for you - personally I like the hands on experimentation - it promotes new ideas as you develop the game and gives a sort of instant gratification that helps drive it forward.
I'm not afraid to experiment (some great gameplay mechanics have arisen from accidents), I'm not afraid to deviate from my main project (I'll often keep several projects of different styles/genres on the go, and flit between them when I'm stuck or bored with one as it helps keep interest in game development high). I think the most important thing is to do it for the love of doing it, not because there is a chance of success or because you're trying to break into the indie scene. Make the game for yourself first and foremost, and if you manage to finish it (much easier said than done) then start to think about a wider audience.
It's hard when you're working a full time job, have family commitments etc to devote time to developing a game - I cut out things that are less important - cut back on TV, block distracting websites, work on problems in my lunchbreak, after work etc. I try to grab an hour or so on most days, but then sometimes go full swing on weekends/holidays. You'd be amazed what you can get done in a short time when you've spent months thinking about and immersing yourself in the project.