Let's say you're a restaurant owner, and you want to see exactly how important good table service is to your business. If you use the NPS system, you will have a number of reviews (possibly many if you own multiple restaruants) with ratings between 0 and 10 and comments.
You categorise each review into a number of predefined categories - good/bad table service, good/bad food, good/bad atmosphere , etc. If you categorise enough of these reviews, then you could fit a multiple linear regression model to quantify exactly just how much good table service affects your NPS score.
This is one way you could visualise these results. You can interpret the above pretty easily - good service adds 30.8 points to your NPS, while a meal that isn't good will take away 45.2 NPS points. You can quickly see how important each factor is.
Note that the values and the categories should not be taken literally - they were simply made up for illustration purposes.
See http://www.puzzlr.org/ for more examples.