Any data set is a lot more useful if you segment it.
As an example, let’s say you find out that your e-commerce website converts at a rate of 3%. That is, for every 100 visitors, 3 make a transaction. That’s somewhat useful data, but it isn’t actionable -- i.e. you can't do much with it -- until you segment it.
You need to break the users into segments: by gender or age or income, etc. When you do, you’ll find actionable insights that will allow you to take actions that will increase your conversions. For example, you might find that men between the ages of 30 and 40 that make more than $100k per year actually convert at the rate of 20%, but that most of your site’s visitors are in lower converting segments, thus the aggregate 3% conversion. With information like this you can adjust your marketing to bring more higher converting users to your site -- you'll get more marketing bang for your buck.
We must do the same with our unemployment data. The unemployment rate -- last time I checked -- was 9%. This number is quoted over and over again in the media as if, by itself, it actually means something. 9% unemployment is not actionable. It must be segmented.
For example, the U.S. unemployment rate for those with graduate degrees is 2%, college grads 4.5%, high school grads 9.7%, non-high-school grads 15%.
It’s critical to recognize the difference between these segments. The data is telling us that for the educated segment of our population, unemployment is at or well below its natural rate. But for the uneducated population it’s super high. This is actionable data. This tells us that there isn’t necessarily a shortage of jobs. There may actually be a shortage of qualified labor. Politicians should keep this segmentation in mind when evaluating "job creation" vs. "job training" programs.