To me, any measure will work as long as it is understood by the people who use it. If you make 300 parts and one is bad, you can call that

* 0.33% defective

* 3300 ppm defective

* 1:300 defective

* 99.67% RTY

* 0.0033 probability of defective

Mathematically, they are all the same.

You could also use "standard" conversions and find a Sigma level or a Cpk number, although these calculations can be misleading, since they are typically based on idealizations like a normal distributions.

Tim F