If we're talking about a keezer, holding ambient steady is all that matters. There's no differential. And I still prefer that approach with fermentation. As long as ambient is all you can affect with heating and cooling, that is, you don't have a jacketed fermenter, I find that when putting the probe in a thermowell or on the fermenter, you will inevitably end up with the ferment itself having temperature swings: It will have already warmed up above set temp before the controller kicks on, and then it will overcorrect, as it will keep cooling running till the liquid hits set temp, when ambient will be much lower, and the liquid will then continue cooling, and your heat will kick on, and here we go again.... I just control ambient, and from experience I know what the differential between that and the ferment will be. (I monitor the ferment with a separate probe in the thermowell.) This works for me, but everybody will figure out what works best for themselves. It all depends on your particular equipment, and, I'll add, your desired degree of precision and willingness to take a hands-on approach.