I've been using this at work for the last year or so to great success.
For example, we have an internal phishing simulation/assessment program, and want to track metrics like improvement and general uncertainty. Since implementing this about a year ago, we've been able to make great improvements such as:
* for a given person, identify the wilson lower bound that they would not get phished if they were targeted
* for the employee population as a whole, determine the 95% uncertainty on whether a sample employee would get phished if targeted
It lets us make much more intelligent inferences about things, much more accurate risk assessments, and also lets us improve the program pretty significantly (e.g. your probability of being targeted being weighted by a combination of your wilson lower bound and your wilson uncertainty).
There are SO MANY opportunities to improve things by using this method. Obviously it isn't applicable everywhere, but I'd suggest you look at any metrics you have that use an average and just take a moment to ask yourself if a Wilson bound would be more appropriate, or might enable you to make marked improvements.
For example, we have an internal phishing simulation/assessment program, and want to track metrics like improvement and general uncertainty. Since implementing this about a year ago, we've been able to make great improvements such as:
* for a given person, identify the wilson lower bound that they would not get phished if they were targeted
* for the employee population as a whole, determine the 95% uncertainty on whether a sample employee would get phished if targeted
It lets us make much more intelligent inferences about things, much more accurate risk assessments, and also lets us improve the program pretty significantly (e.g. your probability of being targeted being weighted by a combination of your wilson lower bound and your wilson uncertainty).
There are SO MANY opportunities to improve things by using this method. Obviously it isn't applicable everywhere, but I'd suggest you look at any metrics you have that use an average and just take a moment to ask yourself if a Wilson bound would be more appropriate, or might enable you to make marked improvements.