I certainly agree that a repeatability study is probably not warranted for most scales used to count parts by weight. However, a MSA is a measurement system analysis which is more than a ‘simple’ Gauge R&R study. The purpose of any control is to ensure the quality of whatever it is assessing. The MSA is intended to establish this ability. If the system can’t achieve it’s goal of assuring quality then you need a different system.
For counting scales the analysis is to see if we can reliably differentiate between 4 parts and 5 parts, or 100 parts and 101 parts,etc. essentially, the weight of the parts themselves has to be consistent enough that you actually can count them using the weight. If you can’t then you need a different system. Perhaps a ‘counting’ sensor can be used or packaging like an egg carton that specifically holds the correct number of individual parts.