Reference sample size 30, reasonable amount to measure / analyse.
approximate 10% margin of error between a sample of 30 to 500.
Of course it would be better to do 500 for accuracy but you must take into account the cost between 30 and 500 pending what you are measuring. Remember the std dev (10%) margin and you should be fine.
As Shree commented, i would not have thought that this would fall under ‘six sigma’ project. It would be better using lean manufacturing techniques. I deploy a mixture, using six sigma tools and lean manufacture together to root out real causes. In cases like overtime, process flow analysis used with activity based costing would be ideal.
To answer some of your questions, regarding return on investment, this can be measured in monies which all understand, also can be measured for ‘culture change’ and increase in a higher trained workforce.
one project is usually enough, for beyond that their approach to day to problems should be that of the 6s approach.
One interesting fact, is that they have ‘quoted ‘ 90% is your problem, this is very significant statement from them . Therefore they must have significant data, the six sigma tools are ok but interpretation can be difficult in wrong hands.!