[Doc]

I generally agree with Slim on multiple instances but there are exceptions though.
For example, a mix I am finishing now has a thunderclap at the end that I want to create some impact. I put an L2 on that channel and pushed it a few dB for added punch without needing to sacrifice the level of the rest of the song. The final master of the album this track is from will probably have an L2 on it too with minimal or no gain.
Something else to consider is whether you want all tracks normalised (maximum level without compression). Quite often the level your ears perceive and what the meters tell you is quite different. This is partly due to the non-linear sensitivity of human ears and the varying frequency content of different songs. My advice is to trust your ears and adjust levels to suit.
Also, it is my opinion and indeed the opinions of most other mastering engineers that less is usually better. Get it right before mastering. The best compliment you can get is for the mastering engineer (or you if it's a DIY) to say, "I don't need to do anything to that".