A colleague of mine showed me this approach. I think the idea behind it is to get an even loading of material across the top of the bed. The approach allows for a more equal distribution of extract on and into the silica particles which should help with chromatography. Drying improves this. Look at ACS publications Gerald Wächter - he may describe the approach.
With a few exceptions, "dry loading" tends to improve resolution. The reason is that the compound can be dissolved in whatever solvent is convenient to the chemist. The evaporation removes the solvent, so it doesn't affect the chromatography. If you are running a hexane/ethyl acetate flash method, you can dissolve the sample in 100% ethyl acetate, add silica, and evaporate the solvent to adsorb the compound onto the silica. With the ethyl acetate gone from the loaded sample, it will elute down the column when the solvent composition is correct. Without dry loading, the solvent prevents the sample from interacting with the silica, causing a loss of resolution.
Liquid loading is more convenient. It works especially well when the loading solvent dissolves the sample and does not cause migration. Frequently, a mix of DCM and hexanes which barely solubilizes the sample will work. Otherwise, I switch to reverse-phase C18 silica and perform the same trick with DMSO/ water mixtures instead.