The other day, I followed a link to a website promising to tell me “the best way to communicate with any prospect, customer, or coworker based on their unique personality.” I was skeptical—and then, once I tried it, I was a little concerned about its accuracy. How could it possibly know about my affinity for emails containing attachments or my dislike of hyperbole? Without actually searching through my emails, at least.
I had stumbled across Crystal, a new service that analyzes publicly-available data sources to come up with personality profiles for coworkers and friends, and then recommends ways to communicate with them. If you pony up for the paid version of the service, Crystal will also offer real-time editing suggestions in emails, tailored to the recipient’s personality.
Crystal was masterminded by Drew D’Agostino, the co-founder of management software company Attend.com. In the summer of 2014, he left that job to experiment with the possibilities of using online content to gauge people’s personalities.
“I naturally don’t have a high emotional intelligence, and I saw a lot of email miscommunication happening,” he says. “So I started building this algorithm to detect personality type. After a few months, it got to be kind of scary accurate.”
Here’s Crystal’s basic summary of my personality:
It’s a decent representation. But you’ll notice that Crystal’s confidence in the answer is only 50%. That’s based on how much data Crystal was able to find about me on the Internet.
According to D’Agostino, the information that Crystal finds is essentially what you’d discover by Googling somebody—company bios, Amazon and Yelp reviews, social media profiles, and so on. Crystal weights certain sources more than others—a big block of text written on a review site gets more weight than a retweeted article, for example.
I was surprised at the low accuracy confidence of the score considering how much publicly-available text I write as part of my job. “For a writer with tons of articles, there’s actually a lot of noise in that. It’s stuff that’s not necessarily talking about yourself,” says D’Agostino. “That can throw it off, but also the algorithm isn’t perfected.”
Here are some of Crystal’s suggestions for communicating with me, based on its personality assessment.
Some of the recommendations are spot-on, but others, like Crystal’s prediction that I enjoy reading instruction manuals and dislike brevity in emails, aren’t as much. Personality assessments can be improved by input from others (Crystal lets users answer questions about others to improve accuracy), but since the service has only been around for a few weeks, I doubt that anyone else provided input for my profile. Overall, I’d rate the accuracy of my profile at about 80%, which is about the average, according to D’Agostino.
Read more at FAST COMPANY