[Bob shapnal]

       When you have the constant equation with the part which is a fraction of the products conc. over the reactants conc. That you use [A] ... [D] for, I know you don't include the concentration of water if that is one of your products or reactants, as it's concentration change, is negligible.

       So you leave it out because it would be multiplied by 1 anyway. But I always don't bother multiplying or dividing by 1000 for the question as fractions don't need to be. However, when water is cancelled out do you have to keep original units because it would mess up the answer?

[Meter]

Contrary to what is taught in high school, the equilibrium constant K (and likewise the reaction quotient Q) are unitless. It is implicitly understood that all of the concentrations [A], [D] ... etc. are all divided by something called the "standard concentration" which is equal to exactly 1 M. This cancels out all units you would get from [A], [D] ... etc. and thus K (and Q) are unitless.

And this isn't some trivial formality, either! Certain equations involve that K or Q place them inside logarithms, and units inside logarithms is undefined. An example of such an equation is [tex]\Delta G^o = -RT\ln{K} [/tex] So in short, you don't need to worry about units at all, and as you mentioned, only reactants/products in the gaseous (g) or aqueous (aq) phase are accounted for in the mass action equation.

Regardless, if you have a factor with identical units in both the numerator and denominator, they would cancel out either way.