"Columns are won and lost on the load."
So very true.
For instance, do you take 25-30 g silica / 1 g crude product literally?
No. Use more for a tougher separation, you can get away with less for an easy one.
If it took more solvent than I thought to dilute my sample do I pick the wider column to thin the band?
Do you mean dilute or dissolve? If you accidentally used more solvent than necessary, then you should concentrate the sample and start again. If the minimum volume required to dissolve the crude is too much for the column size, then you could use a larger column but that's probably not the best approach. A better approach is to preadsorb the crude onto a supporting matrix (silica, Celite etc.) and dry load the sample. Measure an amount of dry supporting material that is about 5% the volume of the column, tip it into solution of the crude and concentrate. This should give a free flowing powder that you can tip onto the top of your column. If the crude/support mixture is sticky or lumpy, redissolve it, add a bit more of the supporting material and concentrate again. The total volume of the dry load should not exceed 10% of the total column volume (unless the separation is extremely easy).
Another alternative is to use a different load solvent - one in which the crude is more soluble. You need to make sure that the solvent is appropriate and will not drag your material too much on the first flush. TLC your crude using your proposed load solvent as the mobile phase, if the compound of interest stays close to the baseline you are good to go. Toluene, DCM and chloroform are usually good candidates.