Numbers like this are often manipulated in some way. Sure, 99.9% of students who failed school last year did not fail it this year, but what about the 30% of them who transferred out? 99.9% of students who retook the test improved, but maybe only students who failed were asked to retake the test. 99.9% of teachers hired last year like the school district. Manipulating statistics is an art form, and it's usually in the hands of those who want to prove their point or sell their product. That's why it's surprising to hear an actual customer say that a product fixed a problem at a rate of 99.9%.
I help my friend show people how effective his Cranium cover/case/protector can be on Chromebooks. I've heard some decent feedback, but this one took the prize. A school technology guy who is testing the Cranium at one middle school sent a message that the breakage rate is down 99.9%. He might be overly excited, but the point is that it's not my friend throwing numbers at me from his "independent" tests. This is the opinion of the customer, who has probably seen near-20% breakage rates drop to below-5% (maybe below 1%). This may not constitute 99.9%, but it's probably an exciting change.