19 Jun 2017

Can Neuroscience Find You the Perfect Job?

Pymetrics CEO and cofounder Frida Polli matches cognitive traits to career paths


Frida Polli (MBA 2012) is the CEO and cofounder of pymetrics, which uses neuroscience-based games to help match people to their perfect careers. In this episode, she talks with Bulletin editor Dan Morrell about which characteristics match with which career paths.


When we talk about a career path, we often talk about it as a journey of discovery. You go out. You try a few jobs. And if the stars align, you gradually find something that suits you best.

But what if that process was flipped? What if you didn't have to wait for the world to tell you what job you're best suited for? What if you had the answers before you started your career?

That's the premise of the career-matching firm pymetrics. Series of neuroscience-based games, users can determine their unique personality traits, match them to their ideal career path. It works for recruiting too, allowing companies in search of specific personality traits to find ideal workers.

Today I talk with neuroscientist Frida Polli, cofounder and CEO of pymetrics, about what characteristics make for, say, a great consultant or a venture capitalist and why she thinks no one is unemployable.

Dan Morrell: Frida, your company pymetrics employs a series of neuroscience tests that help determine personal traits and characteristics. I assume that you took these tests early on in the development of the company. What did they tell you about yourself?

Frida Polli: I did, Dan. I actually did take the test. And what was very interesting is that at the beginning of pymetrics we didn't have any career matches.

The way we develop a career match is we have lots of other people play the games, and then from the entire pool of people that have played them, we select individuals that are good representations of different careers. We group them all together, and when we have enough of those, we will build a career profile.

So when we have enough consultants go through it, then we'll build a consulting profile. So when Julie and myself went through pymetrics, there were no career matches available because we hadn't built that side of the platform. So we learned interesting things about the general cognitive and emotional traits that we all have as human beings but not the career matching.

And, interestingly, I found things like I was very altruistic on the altruistic side of the spectrum, which was interesting to learn. I also learned things about my attention, which I kind of already knew but were confirmed for me in the sense that I tend to be somewhat impulsive. So those pieces were interesting.

What was even more interesting is that, then, once we had enough data to build a lot of the career matching profiles, we got our career matches. So we were all very excited when the day arrived that we could see what am I really a good fit for, because it could tell me that I'm actually a really good fit for something that I'm not currently doing.

It was kind of a highly anticipated moment. And as I always say, luckily for me, entrepreneurship came up in my top three matches as did STEM research. So STEM research is what I used to do. I used to be a scientist.

And it also confirmed things about myself that I suspected to be true, which is that being a consultant or being a lawyer would be very bad career matches for me. And I had always thought that about myself, but it was confirmation.

Morrell: So at what point in the development of the company did you know that this was going to be a company?

Polli: Well, Julie and I always had lofty goals for ourselves. So we always thought it was going to be a company. In fact, we always considered ourselves a company even when it was just two people and we weren't getting paid. But we had a company, or the makings of a company was when we met our investors.

So we met Khosla Ventures back in late 2013. And they're, in our minds and in other people's minds as well, very good investors, specifically when it comes to science-based companies. And so really this was, to my mind, kind of the be all and end all. I just really admired a lot of the companies they invested in. I thought very highly of them.

And so I sort of approached the investment process with some degree of skepticism in terms of how interested they would be, and it was a very fast turnaround. They basically decided in a day to invest in our seed round, and it was-- that was probably the first signal of we were on to something here.

Morrell: Does your insight into how venture capitalists think help you in those discussions at all?

Polli: That's a really good question. We did an analysis of venture capitalists versus entrepreneurs and the similarities and differences, and they had some interesting similarities. But one of the differences-- you remember how I told you about altruism?

Morrell: Yeah.

Polli: So it turns out that our entrepreneur profile actually indicates-- and I've seen this in other research before-- that entrepreneurs tend to be on the more altruistic side of the spectrum. Venture capitalists are not. But certainly, I think, in thinking about the VC profile, I think even if you have a cool technology and you think it's going to change the world, I think you have to make a business case for it. But I think sometimes potentially entrepreneurs get carried away by how cool their technology is and not what is the big market that you're going to address.

And so always just keeping an eye on what is the bottom line. I think that was kind of helpful in thinking about pitching investors and, again, backed up by the data that we had.

Morrell: One of the interesting things here is to understand what potentially could make a great consultant, what potentially could make a great entrepreneur. What made me take the test and say that it was 90% sure I was a writer? What are those things?

Polli: We are constantly going through the data and discovering things that we didn't realize, right? And I think one of the things was what we discovered about salespeople, which is that they tended to be impulsive and inattentive. And that was kind of potentially counter-intuitive.

Maybe you just think, oh, that's a bad trait to have overall for any profession, right? But it turns out, hey, this was linked to high-performing salesmanship, right?

And so then the theory behind that could be that we know that people that are somewhat inattentive tend to be more creative. So maybe that's why, because there's a creative element to the sales process. People that are inattentive tend to be more novelty seeking, so maybe that's why. They're always on to the next thing.

There are things that we have verified with our data, such as accountants and consultants tend to be very detail oriented. OK, well, we didn't need pymetrics to tell us that. But then there have been these sort of interesting results, like entrepreneurs being altruistic or salespeople being inattentive and impulsive, that I think are more-- they're more aha moments for us.

Morrell: One of the issues that pymetrics looks to correct is bias in hiring. What have you learned about how bias manifests itself in the hiring process? What does that look like?

Polli: This was not something that we had a priori thought, OK, we're going to fix this. It was something that we learned along the way. So what I mean by that is there were many-- there were several instances where I realized, OK, bias is going to be an issue, and we have to correct it.

So the first is when companies identify top performers for us, right? And a lot of times we work with finance companies or technology companies where, unfortunately, their employee set is not very diverse, and there are top performer set may be even worse.

So, essentially, you're looking at a fairly homogeneous top performer set, and again, if we want to just think about the trend, it's usually males and oftentimes Caucasian males. And so that was the first time we kind of thought, OK, well if we're just profiling a very limited demographic sample, it could be that the algorithms will then perpetuate this bias.

That was the first time we really thought to ourselves, OK, we have to-- we really have to look at this. And as we did, we realized, hey, there's some ways that we can use artificial intelligence to correct for this. And it's a really powerful tool in terms of actually calibrating the algorithm so that even if they're trained on a non-diverse set, they will spit out equal outcomes for men and women, for people of different ethnic backgrounds. So that was reassuring. But that was kind of the first instance where we realized, OK, we really have to think about bias.

And then the more companies we've interacted with-- and even becoming an entrepreneur was huge eye-opening experience because-- I think in science it's not 50/50. Business school isn't 50/50 either when it comes to men and women, but it's maybe 30-- maybe it's 1/3, 2/3. You don't feel that much of a skew.

When you become an entrepreneur, a woman entrepreneur, you basically are entering the single digits club, and it's a shock. It was a shock to me to go to meetings or dinners or whatever the case may be and just constantly be surrounded by only men.

It was this interesting parallel of my-- the technology company that we had started, realizing, wow, there really is bias in the workplace and then being an entrepreneur and realizing that also kind of in a much more personal way. It's interesting how the technology and our own personal experience has sort of intersected in that way.

And I have a really funny and-- well, funny's not the right word but a really eye-opening story. I was actually telling this today. So right out of HBS, we were non-funded, working out of a makeshift office. And one of my business school classmates said, hey, I'll be your intern for the summer.

So he came to intern with us, and at the time we were doing business development, someone had set us up with a meeting at a bank. So we go to this meeting, and I clearly present myself as the CEO and founder, and I present my male classmate as my business development intern.

And the banking person spent the entire time-- I'm not kidding you-- speaking fully to my classmate. And it was a little bit off-putting for him too because he kept redirecting the conversation back to this is Frida. She's the founder. I'm sure she could explain more. And he still quotes it to this day. He still tells me, like, Frida, I didn't really believe that sexism in the workplace existed, and then we went to that meeting, and it was just shocking to me.

And so we've had a lot of interesting experiences since we started pymetrics, and I think it's just brought home the idea that there really needs to be some bias correction going on. Can I give you another example?

Morrell: Yeah, please do. Yeah.

Polli: I can't tell you the number of times I go to give presentations where-- I'll be dressed up in my suit and everything else, and I'll be waiting to go to the podium. And some other speaker will come to me and be like, oh, where's the clicker for the presentation slides? And I'm like, well, I'm not with the event. I'm following you as a speaker. Again, people don't realize this, and they don't-- and then they'll come and apologize to me afterwards, and they'll say, oh, I'm really sorry.

And so it really has nothing to do with people being badly intentioned. It's literally, if you don't see a lot of women as entrepreneurs, as business leaders, you just don't expect that when you see them in that context that that's what they are. And so I think that's really what pymetrics is trying to change, is really, from a algorithmic perspective, really trying to put more women in those role-- women and people of non-Caucasian background in those roles.

Morrell: So it didn't necessarily start out with that social mission or outcome--

Polli: No.

Morrell: --in mind, but it has had that impact.

Polli: Yeah. It didn't start out with that mission, and I wouldn't even consider it a social mission, Dan, because what I really-- it still sort of aligns with the original mission of making hiring more scientific, which is-- quite frankly, you're making an unscientific and unsound decision if you're preferentially selecting one group of people over others. It's actually leading to you selecting people that are worse performing.

And there's an article that was published a couple years ago by a group at MIT that showed exactly this, that, unfortunately, because of our biases, we're more likely to hire somebody who is lower performing but of a demographic that we expect in that role rather than a higher performing person that's not in that demographic. It happened to be men and women in this case, but it's really fascinating. So it's not about a social mission like, oh, we should do this because it's out of the goodness of our own hearts.

Morrell: Right, right.

Polli: It's really you're actually hiring worse people because of your own biases, and I think that message needs to be driven home a lot more. And there's nothing that irritates me more than people that are like, oh, I'd love to hire a woman or a minority, but I want the best candidate. And you're like, actually, it's the inverse statement. You know what I mean?

You should be saying, like, I want to hire a woman and/or a person of minority because they're the best candidate, not instead of or in spite of or somehow making it sort of like an antagonistic relationship. It's actually not at all.

Morrell: That brings up a related question, which is sort of more general than that. I wonder, do you think business has trouble sort of trusting science in some general way?

Polli: I don't think I would call it a trust gap. I think that AI has been not well portrayed in the news necessarily. And I also think-- I think, unfortunately, the media likes to take a clickbaity spin on a subject. Oh, AI, like the evil overlord robots are coming to steal your job, right?

We were at TED last week, and we actually gave an audience response to some of the speakers that were presenting AI-related stuff at TED. And we actually got to present on the main stage. And the whole pitch that we gave was AI is an amazingly powerful tool for, instead of creating evil robot overlords, to actually democratize the world in a way that humans, because of their biases, are not able to.

So I think a lot of it is what's written up is biased-- no pun intended-- in sort of pitching them in a certain way. And then I think it's like any other profession where there's been a huge influx of technology-- this happened in marketing about a decade ago-- that people that are in that profession are sort of overwhelmed.

And they don't really know what to make of all of this technology. And so they're rightly so, I think, doing their diligence and figuring out, OK, what-- which aspects of these-- what aspects of this technology is really going to help me versus not necessarily do that?

So, again, is there a slight distrust factor? I think, yes, there is, but I think those are some of the reasons behind it. I'm seeing it daily moving. A couple years ago, it was really hard to have a conversation, and now I think people are seeing the benefits. And they're like, wow, I really-- I want me some of that, you know?

Morrell: Frida, how many people have taken the pymetrics test?

Polli: We have clients using it globally all over the world in fairly large amounts, so I think it could be as high as half a million. It's in the hundreds of thousands for sure.

Morrell: So you have all this data about people.

Polli: Right.

Morrell: What has it taught you about how we think?

Polli: This is really, I think, one of the fundamental takeaways. Whenever you mention the word test or assessment or something, we are so conditioned to think about those things as being good, like telling you that you're good or bad, right? I think it's just-- school drives that home. Standardized testing drives that home.

And from the beginning, pymetrics has always been around, no, it's about finding your fit. Basically, no one's unemployable. I can't tell you the number of times I've talked about pymetrics, and the joke has always been, oh, my God, I'm going to take it, and it's going to tell me I'm good for nothing, right?

So what it's taught me is that that's actually true, that people range greatly in their cognitive and emotional traits, which we knew ahead of time. The patterns that make for success in different fields are so varied, and I think that it's really sort of-- it's really inspiring to me, in any case, to realize that it really is about every pot has a lid. And it's more about finding where on the stove you sit than it is determining that there are certain people that are just superstars in everything that they do and other people are not.

And the thing that I feel the proudest about is that it's really about helping anyone, whether they are-- whatever they think they may be good at and whatever their educational background is or whether they've had a learning disability in the past or whatever it is, really figure out, hey, I'm unique, and my unique traits make me good for these types of things. And there's no version of uniqueness that isn't-- that doesn't have there fit somewhere in the world of work.

Skydeck is produced by the External Relations department at Harvard Business School and edited by Craig McDonald. It is available at iTunes or wherever you get your favorite podcasts. For more information or to find archived episodes, visit alumni.hbs.edu/skydeck.

Featured Alumni

Featured Alumni

Class of MBA 2012, Section D
follow @fridapolli

Post a Comment