Episode 013: AI-Enabled Personality Testing with Drew D’Agostino

In this episode of the SaaS CX Show, I’m talking with Drew D’Agostino, founder of Crystal. We discuss the role that AI plays in his software and how personality tests are such a valuable tool for businesses.
Show NotesTranscript
AI Personality Testing in Crystal with Drew D'Agostino
Drew D’Agostino

When most people talk about artificial intelligence, it’s usually in reference to managing massive amounts of information and distilling it into a digestible package – not about personality tests. Amazon uses AI to understand its customer’s needs, Microsoft uses AI to develop more responsive software – the list goes on.

However, while AI is usually reserved for making technology better and more user-friendly, it can also be a tool for bringing people together. Such is the goal of Crystal, a machine-learning program that incorporates data and personality tests to help people understand each other better.

In this episode of the SaaS CX Show, I’m talking with Drew D’Agostino, founder of Crystal. We discuss the role that AI plays in his software and how personality tests are such a valuable tool for businesses. Here are a couple of highlights from the show.

Personality Test vs. Data Mining

Typically speaking, personality tests are inherently limited. You can take one, you can have your team members take one, and you might be able to get those within your inner circle to take one. However, if you want to facilitate the widespread adoption of a test, how can you go about it?

Crystal is revolutionary because it merges the standard of test-taking with data mining from a variety of sources. The program looks at social media profiles, resumes, LinkedIn pages – all to create a more comprehensive understanding of a person. What makes this process even more exciting is that it’s still really accurate. According to Drew, AI and data mining yields about 80 to 85-percent accuracy, compared to 97-percent when someone takes a personality test.

So, why does that matter? Well, businesses can get valuable information about clients and potential employees long before they meet in person. Rather than having to discover the personality of these people organically, they can facilitate stronger interactions right off the bat.

Want to use video to communicate? Check out this interview

Artificial Intelligence as a Human Tool

As we mentioned, most people talk about AI as a means of digesting big data or as a way to make technology more personable, not as a personality test. With Crystal, however, Drew and his team are trying to help improve the way we communicate. It’s a tool for human-to-human interaction, not human to machine. As Drew puts it, that is the company’s north star, and it’s the reason he loves coming into work every day. Enabling users to find better communication methods and facilitating stronger relationships is what motivates him.

So far, Crystal’s success shows that there is a lot of potential in this software, and we’re excited to see where it goes. We talk more about his process and the value of Crystal during the show, so be sure to check it out here. You can also find out more about Drew and his program at www.crystalknows.com.

Frank Bria (00:00):
The SaaS CX Show, Episode 13

Announcer (00:04):
From founders and CEOs to founders and CEOs. It’s the SaaS CX Show. You’ve found the one stop shop for all things CX. Each interview is an in depth analysis of a successful growing SaaS company, building a world class customer experience for their customers. And now for your host, serial SaaS entrepreneur, founder, consultant and advisor, Frank Bria.

Frank Bria (00:31):
Hey everyone. Welcome to the SaaS CX show by SaaS founders and CEOs for SaaS, founders and CEOs. I’m your host, Frank Bria. Today’s episode, we are going to talk about personality typing, but first, the SaaS CX Show is brought to you by the SaaS CX group. The number one reason for ideal customer churn, not getting value from your software. Find out how to fix that by downloading our SaaS churn checklist. We covered the seven things you need to have in place to ensure your ideal customers stay, use and love your software. Decrease churn by 10 to 25% just three, six months. Find out more at the show’s homepage. Saascx.Show. That’s saascx.show. And now I’m pleased to introduce today’s guest. Drew D’Agostino is the CEO of Crystal, the app that tells you anyone’s personality using AI. Crystal accurately identifies a person’s motivations, communication style, and other behavioral traits. Thousands of professionals globally use Crystal to communicate more effectively, write more persuasively and build trust faster with new people. Previously, Drew was CTO of attend.com an event management software company. Drew and his company had been featured in Inc, Fortune, CNN, Fast Company, MIT Technology Review, Wired and The Guardian. Drew, welcome to the show.

Drew D’Agostino (01:50):
Hey! It’s good to be here, Frank.

Frank Bria (01:52):
Absolutely thrilled to have you! So first of all, personality typing as an app. That’s just kind of a new take on this. Can you walk us through a little bit what Crystal does?

Drew D’Agostino (02:04):
Yeah, sure. So most people are familiar with some kind of personality type. A lot of people have just encountered either like Myers-Briggs or Disc or Enneagram, it’s a really popular one now, or the Big Five. So this is become a, you know, kind of a pop psychology thing, but really under it, there’s a lot of really interesting and relatively new science as well. So basically what Crystal does is we try to discover or analyze whichever personality framework you want and determine what someone’s type is so you can communicate more effectively. And we try to do that in as real, convenient, and quick way as possible. So that means that we have these personality assessments in Crystal, so like Disc tests and Enneagram and Myers-Briggs types tests. And those can help companies basically get personality types for all their employees for free. And we also offer personality predictions, which is using machine learning to analyze social media, text samples, resumes, any other meta-data about somebody that can be correlated with personality types. So the goal is to get an accurate idea of somebody’s behavioral tendencies and you know, whatever situation you have, whether you’re meeting with them in a sales meeting or you’re talking to them in a one-on-one for management purposes, you know, so that’s our goal.

Frank Bria (03:33):
Yeah. What’s fascinating to me is I was kind of looking over what you guys are doing is that you’re running the person, in some cases it looks like you’re running the personality assessment without the personality test. You’re looking at social media. I think you have a resume, a connector. Is that predictive?

Drew D’Agostino (03:55):
Yeah. So we do have personality tests in Crystal and we have thousands of people that sign up and take regular personality tests each week. So that is kind of like building a big training set of data. And that’s also always the most accurate way to get a personality type. Something like 97% validity. However, there are a lot of situations where personality tests are not feasible or they’re not economical or viable mainly when you don’t know a person. So think about like personality data has been always kind of limited to who you can get to take a test. So that usually means like just myself or somebody I know immediately or my team that I am managing them. I can kind of force them in for a big session and we can hire a consultant. They can take a personality test. So it’s this like very limited data set but it’s really valuable because when you notice how somebody tends to think and behave, whether that’s from knowing them for a long time or whether that’s from a personality assessment, it really helps you a lot. It’s like this whole lens of looking at that relationship, cause it’s a very great way to get to know somebody on a deeper level and really understand what motivates them and how you can speak more effectively. So by not requiring a test and by just doing the best we can to analyze what data we have, we can achieve somewhere around an 80%. They kind of always fluctuate between 80 and 85% accuracy and that is based on a lot of, five years of data that we have. Or if we have enough, say like call it a large LinkedIn profile, we can properly guess where somebody falls on one of these four major personality quadrants. If you’re talking about like Disc and tell you, okay, here, this person’s really likely to be very direct and to the point. So if you’re on the phone call with them, you want to get, you know, kind of skip the small stuff, small talk, don’t go into your story about your dog, give them the facts right away. And again, that’s only gonna be right four out of five times, but four out of five’s a lot better than, you know, one in five. So those are the kinds of situations where this kind of predictive or we call it personality AI really comes into play. But when you have a team of people that you’re managing, assessments are still the way to go because those are more accurate, more valid.

Frank Bria (06:15):
Yeah. So, I mean obviously if you can do the test it works better. But it sounds like what you’ve done is your algorithms are kind of taking the the information where it is a lot more accurate, where you’ve got the tests and I guess what, you’re back testing it with the social profiles of the folks who’ve actually taken the test and kind of matching, you know, getting the algorithm to learn from that essentially?

Drew D’Agostino (06:36):
Yeah, and other information, we collect other data from Crystal. So for example, you take, there’s not just one personality test, you can take a Disc test, a Myers-Briggs test, Enneagram, values, motivations, strengths. There’s a lot of these different assessments. So we can map, say your Disc test, to your job title and your resume or whatever else we have access to. But we can also map personality tests across each other. So find out which Enneagram types or which Disc types, we can do predictions in that way as well. So it’s this interesting dataset that we are still building.

Frank Bria (07:17):
Awesome. That is fascinating. Now, the interesting thing is I think a lot of people have the impression though that social media is sort of a facade. You know, what people put out there is kind of what they want people to think they’re like, but deep down, that’s not really the case. Is your data indicating any of that or is your data indicating that we might think it’s a facade, but if you dig deep enough, actually the real personality is sort of sitting in there someplace.

Drew D’Agostino (07:44):
That is where the 20% comes from, the 20% inaccuracy. So that’s, that’s, yeah, so it’s definitely a thing and there are times when you’re looking at someone’s social media presence and you can kinda just tell that they are aiming for a certain tone or style and Crystal picks that up. I mean, Crystal is not magic. It’s looking, it’s basically saying somebody who writes like this, typically gets this score on a personality test. So yeah, that definitely happens. It does happen, but not a majority of the time, so.

Frank Bria (08:17):
Yeah, so 20% of us are sociopath’s basically

Drew D’Agostino (08:24):
Special personality type for those I think in a big, red page. Please feed possum(?)

Frank Bria (08:31):
That’s awesome. Again, as you are taking the technology forward everyone has sort of looked at personality profiling as a stepping stone to something, right? So selling better or communicating better or coaching your employees better, having team members get along better and when you just do the personality profiling, it’s almost like you learn all this stuff and it’s all theory in your head, but you’re sort of left on your own to execute this. You guys have put some tools in place to actually make some of this stuff super practical. Can you walk us through some of the use cases where you’re actually leveraging this to help professionals?

Drew D’Agostino (09:13):
Yeah, that’s a really good point. That’s always a big question. It’s like, all right, now that I know this about someone and know that they’re a, you know, ENFP or a DI or whatever. What does it actually mean? What do I do with it? So we’ve put together a pretty full suite of, I call them coaching tools that are designed to take that personality bit and then translate it into ROI. And there’s a little bit of a challenge for us because there’s so many different types of people who use Crystal. So you’ll find on there tools for salespeople. So there’s like a coach where it says, all right, what do I need to do? I’m getting on the phone with this person and I need to talk to them about pricing. Cause that’s gonna be very different based on the personality. You know, if someone wants to get into details of the pricing, someone’s probably just gonna want to understand the story and get social proof. So there’s, it’s a very big deal. So that’s really useful for salespeople. But then there’s other parts of Crystal where if I’m managing a team or building a team, I can create a group report and take these six personalities and see where we’re lacking or where we might have weak spots or potential conflict. So there’s all these in- between the coaching reports, different types of insights that you can do on profiles. There’s lots of things that you can do and it really varies based on the job you’re trying to accomplish. It’s an ongoing challenge for us to try to, you know, everyone says you shouldn’t make a product that’s everything to everyone. But every time we really tried to focus on one audience, we end up getting dragged back into this more general area. You know there’s a lot you can do. You’ve just got to explore the product for yourself and you can do most of it for free.

Frank Bria (11:02):
Well, you’ll just leave that to your product manager to manage your roadmap for you. That’ll be the toughest job ever. You guys are looking at there are a lot of overlaps between these use cases though. So I totally understand what you’re saying. You know, one person is probably going to be looking at a number of different communication cases. This leveraging of AI in in communication is really fascinating because most of the communication use cases in AI started off with one person sort of interacting with the computer, whether it’s a bot communication or whatever, but you guys are leveraging AI to improve the communication between two human beings, which is really fascinating. What do you see as the next step for you guys in leveraging more powerful algorithms, better data as you’re growing this dataset? What problems are you going to be able to address that we can’t now because we just don’t have either the computing power or the algorithms or the data?

Drew D’Agostino (12:06):
This is a really fundamental question that kind of lies at the heart of what we do. And it’s my existential crisis like every week. Cause at the end of the day there’s a lot of buzz and hype around AI and also a lot of money following it. So it’s very hard to stay on track. So you really need to have a clear idea of what you’re aiming the business at. So our North star with Crystal and what we’re trying to do every day, is we are trying to make our users more confident, more effective, better communicators. We are not trying to replace the need for our users to communicate or to do the emotional legwork to do whatever kinds of you want to call it. We’re not trying to just accomplish the task for them, we’re trying to level up their skills. So that really affects what we do with our product, right? Because there was a time when we had this email tool, which was one of our more popular features where you could use it to correct your emails, kind of like Grammarly thing. And you can still use this in parts of Crystal, and the pull of that product, if we were not really focused on making our users better, more proficient communicators, the pull of that product would actually be how much of this email can we write for our users? And there are companies that are trying to do that. Because of just the core focus of our company and honestly just what I want to be working on every day, I don’t want to be building robots that take the empathy or the humanity out of online communication. I think there’s plenty of that. What I would rather be working on every day is, all right, how do you use technology to accelerate relationships between people? The fact that I can look you up before meeting you for the first time and understand not completely about you, but have this extra insight that’s going to help me connect with you quicker, that to me is, it’s a very human use of technology that I hope that, actually, I view Crystal as at least like a 10 year project for me. I’m five years in. I wanna be working on something that actually I feel good about, so.

Frank Bria (14:28):
Well, it’s interesting because what you’re describing is AI not as an efficiency play, but as an effectiveness play in the communication where I think a lot of people do, as to your point, are focusing on the efficiency elements, right? Is technology accelerating the process or taking out tasks, making things faster, making more money, whatever. But the effectiveness of communication, that’s almost a different dimension of the problem. I think it’s an interesting dichotomy you’ve drawn there in what you’re doing. So that’s great. Drew, this has been a great conversation, unfortunately, we’re out of time. You’ve been super gracious with your schedule. I know you’re really busy, so thanks so much for spending time with us. One last question before we go. People want to learn more about Crystal, what you guys are up to, what’s a great place for them to start?

Drew D’Agostino (15:18):
Just go to crystalknows.com if you want to try the personality prediction tools that we’ve been talking about, you can just add the Chrome extension. It’s really easy from the website and you’ll get 10 credits for free. If you want to have some extra free credits, just shoot me an email and mention the show and we will add some more credits to your account so you can extend your trial. But yeah, it’s a lot fun to experiment with.

Frank Bria (15:45):
Thanks. That’s a generous offer. Thanks a lot. So the link’s in the show notes, if you’re out and about listening to the audio, come on back to the show notes page and connect with Crystal and Drew. Thanks so much Drew for being with us, really appreciate you taking the time.

Drew D’Agostino (15:59):
Great, yeah, it’s been great Frank, appreciate it.

Frank Bria (16:01):
And thanks so much for being with us on the SaaS CX Show. I’ve been your host, Frank Bria. Just a reminder, if you want to reduce customer churn by 10 to 25% in the next three to six months, check out our SaaS churn checklist, which covers the seven things you need to have in place to ensure your ideal customers stay, use and love your software. Download it at saascx.show. That’s saascx.show. We will see you next time. Make it happen. Bye bye.

Share on facebook
Share on google
Share on twitter
Share on linkedin
Share on pinterest

2 thoughts on “Episode 013: AI-Enabled Personality Testing with Drew D’Agostino”

Leave a Reply

Your email address will not be published. Required fields are marked *