If you want to get deep, you should engage in a statistical analysis to determine what the correct sampling is for the process. That involves math, so everyone, most people, some people, just me might just conduct a simple study to determine a threshold. I'm lazy as all get out, but if this were me, I would weigh 50 or 100 bags daily for a week and record all of the weights. How many of those bags are underweight, what is the average overage, etc? Then make some realistic conclusions about how that might translate into a QC check. For example, if 5% of your bags are underweight, the underlying process is too skinny and needs to be adjusted. If you find 0% under, then you could reasonably set a 2% audit rate as a starting point. Then, over time, you can use your results to justify reducing the checks if that makes sense, although not to a point of never having them. A lot of this comes down to what are the consequences of being underweight and how eager are you to avoid the negative consequences. That's balanced by the desire to not give product away and over-stuffing the bags too much. We fresh-pack fruit and we conduct an hourly QC process that includes weight measurements. As our average overage is 7% and the machinery checks the weight already, we could justify reducing that frequency, but since we need to check for decay, grade, etc., we just keep that check in place as it doesn't hurt and the cost involved in a rejection is high.
Anyway, that's my two cents, euros, pounds.
Good luck,
Todd