Traditionally the peak crystallization temperature is used to calculate the activation energy, which I think relates to crystal growth. Some scholars suggested that by using onset instead of peak, we can have an estimate how easy or difficult it is for the nucleation process to occur.