I suggest you have a read of the original article from Clark Still of Columbia University (W.C. Still, M. Kahn and A. Mitra, J. Org. Chem. 1978, 43, 2923-2925) -- or here
is a good summary, which includes the original paper. I've always found that dry packing the column was best and easiest (as per the original paper) but the three critical issues are the silica gel used, the amount of silica, and the Rf difference for the compounds you are trying to separate.
1. You should use 250-400 mesh, anything bigger will not work properly.
2. Try to use only around 30 times (by weight) the amount of silica to the amount of compound. Most people use far too much, like when they run gravity-fed columns. If you have 250 mg of compound to purify I can guarantee that 7.5 g of silica doesn't look like enough but it is!
3. The Rf ratio
is much more important than the absolute difference in Rf. As an example, say you are trying to separate compound A from B. In one solvent system you get Rf values for A and B of 1.5 and 3.0 respectively. In another solvent system you get 4.0 and 6.5 respectively. On TLC the separation looks better for the second system because the Rf "gap" is 2.5 as opposed to 1.5, but that's the wrong way of looking at it. In the first solvent system compound B moves twice as fast
as compound A (3.0 is twice 1.5) , while in the second system B moves only 1.6 times as fast (6.5/4.0).
A bit of work with TLC to find the best system (I like to run through a variety of hexane/ethyl acetate mixtures) for separation with low Rfs will pay off with better separation.
Of course if you can get your supervisor to spring for a chromatotron...