It would not surprise me at all of the people who set such high standards are basically bastardizing the model behind Net Promoter (a survey company) -- their model is basically this:
- All questions rated 1-10, where a 1 is "Where's my shotgun?" and a "10" is "I want to have your babies"
- They ask a bunch of questions, then ask the only one that actually counts towards the score: "How likely are you to recommend this company/service/product to other people?"
- Scores of 9/10 count as positives
- Scores of 7 or less count as negatives
- The final result is a skewed ratio -- all "positives" MINUS all "negatives", divide that into the total number of responses to get the final score. In other words: 100 surveys, 70 pos, 20 neg, 10 "neutral" (8's), would yield a score of (70-20)/100=50%
A score of 50% is considered normal (the bulk of the companies they run surveys for get 50ish). A score of 80% is considered phenomenal.
I individual scores on specific questions (except that last one) don't matter -- just the biggie that sums them all up. I suppose a way to implement that model in, say , the naval checks mentioned above would be "How likely would you be to serve under this Officer again?". If the Officer gets 50-60, that's not bad at all but not amazing. If he gets an 80, Promote him NOW.