Then it's time to gather outside statistical data. Interesting article by Popular Mechanics, regarding measuring the true (not claimed) physical impact on express packages:

http://www.popularmechanics.com/tech...-your-packages

The parcel was shipped a dozen times (we had neither the time nor the budget to make the hundreds of trips necessary for statistical significance), a modest experiment to see how the device performed and gather enough data to draw broad conclusions.

Nice processes. Measurables, repeatables, and reportables.