I ran across an internet service that uses IBM’s Watson to analyze a person’s written text to provide a personality assessment.

Watson is a really, really smart computer. After I entered several of my own works, it was interesting to see that Watson was able to probe beyond the different writing styles I use for different publications and audiences. It used my words to determine what aspects of my personality really come through.

The result? Watson gave a strikingly similar personality assessment for me no matter what text I had it analyze.

I was also pleased to see how quickly Watson worked. For statistically significant findings, Watson needs at least 3,500 words. Analysis of 10,000 words took less than a second.

So, being me, I wanted to give Watson a bit of a stress test to see what happens when you enter a larger pool of words. I didn’t have to look too far; I accessed over 12,000 comments left by our program participants over the last year in our Metrics That Matter training evaluations.

Those 12,000 comments contained just over 200,000 words, but Watson’s analysis came back in less than 3 seconds.

Before we continue, here’s a disclaimer: This assessment was intended to analyze the writing of a single person, not 5,000, so the findings you’ll see below are far from scientific. But you can still make real generalizations—and come to some interesting conclusions—when you look at the characteristics that deviate from 50% … and there are many interesting ones that do.

And what Watson uncovered about our customers was very intriguing. 


The starburst design is one of several ways Watson lets you look at the information it generates. What jumped out for me right away is a stellar rating in “Sympathy”—99%!

This validated rather than surprised me (though it was still quite encouraging to see), because these comments are from people who train their own staff in our Nonviolent Crisis Intervention® program.

On the other side of the spectrum, Gregariousness (4%), Excitement-Seeking (5%), and Cheerfulness (8%) are rated fairly low. Now before you start thinking that our Certified Instructors aren’t talkative or particularly content, I read it this way: Certified Instructors taking our training were taking the training evaluation very seriously, and Watson picked up on that.

Other traits indicate that Certified Instructors are very much professionally motivated. Looking at Self-Enhancement (95%), Self-Transcendence (93%), and Achievement Striving (95%), you get a general picture that the typical Certified Instructor (if there is a typical Certified Instructor) aims to continuously improve.

As I said before, none of this was overly surprising. We know you guys. And now Watson seems to know you, too.

So while these results are not really going to be indicative for any one individual among the 30,000 individual Certified Instructors teaching CPI training, it’s still fun to see what traits seem to be more common (and less common).   
Want to give it a try yourself? This tool is available on Watson’s Developer Cloud page. You can see how Watson analyzes a few famous people (Oprah, Pope Francis, Gandhi, etc.). Hook up your own Twitter account or enter text of your own!