I would look at characteristics of the clients you like working with and see if you can find other clients like them.

Are they technical? Non technical? Focused on a certain aspect of the niche? What kinds of problems do they have?

Anything that will act as a filter so you take work from clients that have a high chance of being good ones.

As far as expanding to other areas of development, if there are adjacent areas available (for example, using CakePHP if you are already using Symphony, or trying Alfresco if you already do Documentum) it might make sense to investigate and do a trial project in them.

Looking at your profile, it looks like you do interesting stuff. Not sure what industries you target (because it isn't outlined on your site), but you could definitely evaluate industries that have 'hard problems' (and money--don't forget the money).