I used to build full divisions but then noticed that this is a very expensive way of doing it if the individual build times of the brigades are different.
Note that the cost per-day is adjusted to reflect the longest build time. If you build INF and ART brigades together, the game will add up the total IC/day cost of each, then divide that by the build time of the longest in order to get the daily IC cost. Net result is that you pay the same IC/days whether you build them together or independently.
The real savings comes when you build BIG TICKET items, like factories, battleships, and aircraft. Rather than building several together, build one, then start the second one a few weeks or months later. When the first one finishes, the increased Practical bonus will move up the completion date of the second, saving a lot of IC/days, which you would have lost if they completed on the same day. I often start several units together, but rotate a few to the bottom of the production queue where they end up lagging behind due to insufficient IC to go around. On a huge investment like a battleship, I've seen the completion date of the second ship advance by as much as 3-4 months; with factories it's generally a week or two, with each additional factory producing a smaller shift in date.
I regularly drop additional brigades into existing units. In the mid-to-late stages of a couple of games, I made massive changes to the entire divisional structure to make more efficient use of manpower (I had WAY more IC than I needed, but not enough manpower to make optimal use of it), by switching from a 3xINF+ART to a 2xINF+2xART setup for 3 of the 5 divisions in each Corps. To implement this, I built "divisions" of 3xART, moved the entire Corps into the same province as the new brigade, and did a lot of Reorganizing. That left me with a modified Corps, plus 3 INF brigades toward creating fresh divisions (or attaching one or two of those to the HQ to form a 6th division). Multiply that by the number of Corps in the armed forces, and it's a hellish click-fest, but at least it can be implemented gradually. The end result is something like a 20% increase in the number of divisions on the map with only about a 5-10% increase in manpower.