I agree with dbonanno1 that you can rationalize failures during the verification process. For example, at work we have a new combination product releasing and I was part of the team conducting verification shipping study. As a precautionary action in the verification protocol that was drafted we put in that any breakage to packaging would not need an investigational analysis and rather we were conducting the testing for purposes of analytically testing the combination product and not the actual mechanical stresses on the container. We actually did have 3 of the vials out of a n=1000 sample size containing the product break, however since the testing was done in an extreme setting that did not mimic the stress the vials would actually see, aka real world conditions we were able to avoid an investigational analysis. In this case testing the worst case scenario allowed us to prove the product can handle the shipping lane while not having to worry about the packaging.