Every search, like, and click leaves a trace — but what do these digital breadcrumbs really say about you? Author and professor Sandra Matz joined host Brett Hendrie to explore the psychology of data and how businesses use this information to build detailed profiles, predicting everything from your personality to your spending habits. How much do they really know, and what can you do about it?
Every search, like, and click leaves a trace — but what do these digital breadcrumbs really say about you? Author and professor Sandra Matz joined host Brett Hendrie to explore the psychology of data and how businesses use this information to build detailed profiles, predicting everything from your personality to your spending habits. How much do they really know, and what can you do about it?
Brett Hendrie: Have you ever stopped to consider the depth of your digital footprint and what it reveals about you every purchase, every app, every website, it's all tracked, analyzed and used to build a picture of who we Are. Companies can tell if we're shopping for a car have recently adopted a pet, or even if we're expecting a child, sometimes before we've told our own friends and family. But beyond predicting our next purchase, could this data offer a window into our inner lives, into our very psyches, and if so, what are the implications for businesses that use it.
Welcome to Visiting Experts, a Rotman School podcast for lifelong learners exploring transformative ideas about business and society with the influential scholars, thinkers and leaders, featured in our acclaimed Speaker Series, I'm your host, Brett Hendrie, and I'm joined today by Sandra Matz, an Associate Professor of Business at Columbia University, where she teaches and studies human behavior and preferences, focused on the hidden relationships between our digital lives and our psychology. Her work is frequently covered in the media such as the economist, New York Times and BBC, and she's here today to talk about her terrific new book, Mindmasters: The Data Driven Science of Predicting and Changing Human Behavior. It's a fascinating, deep dive into the research behind the algorithms that shape our daily lives. And we're thrilled to have Sandra with us today. Welcome Sandra,
Sandra Matz: Well, thank you so much for having me on the show.
BH: Sandra, your research focuses on this idea of psychological targeting, and I think people in our listening audience are generally familiar with the idea that what we do online is tracked and analyzed, but help us understand what it means when you say psychological targeting.
SM: So the way that I think about psychological targeting is essentially the ability of algorithms to decode our minds and psychology. And what I mean by that is these two steps where algorithms can take all of the traces that we generate as we interact with technology, and it could be anything from what you post on social media to your credit card spending to the data that gets captured by you just carrying your smartphone with you 24/7, and translate these traces into meaningful psychological insights so that could range from your personality, your values, mental health. And then, in the second step, not only peek into your psychology, but potentially change it. So, learning something about your motivations preferences gives others quite a lot of power when it comes to influencing your behavior.
BH: It seems like this has been with us for a while. I mean, certainly with the advent of the commercial Internet, 25 plus years ago, we were getting some form of customized advertising. So, what's changing now in the technologies or how the data is being used and analyzed that makes this a really pertinent question for leaders to understand today?
SM: For me, the big difference compared to, let's say, five, 10 years ago is first of all, that we can understand humans in a much more holistic way. So it used to be the case that we capture your data, and maybe it's your purchases, maybe is what you search for, and we take your past preferences and we directly infer what you might be interested in in the future. Think of Amazon, who looks at what you've purchased in the past, maps it against what everybody else is purchasing, and then makes recommendations. The part that I'm excited about is that we can essentially take all of the traces – and that's not just purchases, not just searches, but really anything from your credit card spending to that, again, data that gets captured by your smartphone – and translate it into something that makes sense to us as humans. If I talk about my friends, if I talk about my family, I don't talk about them in like these segregated traces of behavior. I talk about them as well, there's this extroverted cousin, and then there's my open-minded friend that is just kind of traveling around the world, and that's how we make sense of ourselves and our identities and other people. And now, with the help of machine learning and AI, we can use traces that we generate as humans and translate them into these psychological constructs.
BH: And I think it touches upon an idea that I took from your book that was really interesting, which was the difference between what you call identity claims and behavioral residue, and these are very different types of data footprints that we leave. Can you help our audience understand the distinction between those two terms?
SM: Yeah, so identity claims are all the signals that you very explicitly put out there about how you see yourself and how you want to be seen by the world in some way, right? So in the context of data, that's oftentimes social media. So social media, we paint this picture of our life, sometimes in a slightly self idealized way. We're all slightly happier, more extroverted, maybe a little bit less neurotic than we usually are, and but that's a very explicit and intentional signal to the to the outside world. Now the behavioral residue part is far more implicit and unintentional. So those are all of the traces that you generate, really without thinking. And that could range anywhere from you just carrying your smartphone in your pocket, pretty much 24/7 right? And that smartphone captures your GPS records, so it knows exactly where you are. It probably also knows who you're hanging out with, because that's phones showing up in the same location, oftentimes repeatedly. Those are not traces that you think about. So you're not kind of you're not telling your GPS sensor to track you, and still, there's someone almost looking over your shoulder and knowing what you do at any given point in time. So this is the type of behavioral residue that we oftentimes have far less control over.
If you think about Meta, they get both the explicit identity claims because they see what is it, what are the things that you watch, what are the things that you post about your life, what are the pages that you follow. But they also are on your phone, so they also oftentimes tap into something like your GPS records, your call logs, your gallery. Just take the last app that you've downloaded. Most of us just give permission for apps to tap into the microphone, our photo gallery and our GPS record. So these, especially the big companies, have access to many different data points.
BH: You, in your research, have really focused in on how companies are building not just what you're interested in, but what your psychological profile is. Can you help us understand how accurate are the psychological profiles that are being built by these companies of us as consumers compared to what might be developed in a more traditional setting by psychologists?
SM: To me, the interesting part is that it's not, it's not rocket science, right? Oftentimes, when we talk about these predictive algorithms or AI, we think of it as like this black-black box model, and it's so complicated that we can't understand it. When you look at these relationships that are underlying the predictions, some of them are incredibly simple and intuitive. So if you think of what predicts whether someone is extroverted – they probably talk more about the social activities that they engage in. They spend more money on going to bars. They have a much more active lifestyle when it comes to their physical location tracking. So some of these relationships are really obvious and intuitive. The interesting part is – and that also adds to the accuracy that I can talk about in just a second – some of these relationships are actually something that I might not have even intuited as a psychologist studying these behaviors. So one of my favorite examples in that space is the use of first person pronouns, like “I, me, myself,” which, when I first learned about these findings, I assumed that it was going to be narcissism. My assumption was, well, if someone talks about themselves all the time, that's probably an indication that they're a lot more self-focused than other people. It turns out to be an indication of emotional distress. So when you talk about yourself, you can imagine the last time, if you think back to last time you felt blue, sad and down, you were probably not thinking about how to solve the world's biggest problems. What you were thinking about is yourself. You think about, why am I feeling so bad? Am I ever going to get better? And how am I going to get better? These thoughts creep into the language that we use.
And again, this is nothing that we would have necessarily intuited, even as psychologists studying these topics, and that's what kind of gives algorithms this extra edge over prediction. So when you look at the accuracy, of algorithms predicting your personality traits based on information that you can find on Facebook – so in this case, this is a study that was done on the Facebook pages that people follow. So back to our distinction. Those are very explicit identity claims, right if you follow the page of CNN or the page of Lady Gaga – this is a signal to yourself in the outside world: Here's the person who I want to be. And what my colleagues back at Cambridge University showed is that giving an algorithm access to just your Facebook likes actually makes them better at predicting your personality than people in your environment that should know you pretty well, so that includes co-workers, that includes friends, that includes family members. So those are people who spend a lot of time with you and actually should know you pretty well, and yet, an algorithm, just by looking at your Facebook likes, is better than all of these people. And then back in the back in the study, I think the only person that outperformed the algorithm was someone's spouse. But you can imagine, if you have access to more than just your Facebook likes and more sophisticated algorithms than 10 years ago, an algorithm would probably also, by now, outperform our most significant others. So they're not perfect these algorithms, and I think that's important when you think about applications, but they're pretty accurate.
BH: So you've given us a great insight and understanding into how these systems and this analysis of the data can make an assessment of our psyches and who we are, help us understand the next step and how is this information and this data being used to actually shape our behavior and influence our decisions?
SM: Yeah, and for me, in a way, that was always the almost more interesting step. From a scientific point of view, it's incredibly interesting that you can peek into someone's psychology by observing their data. But if you think about applications, and if you think about impact on society, the second step of once I understand who you are, your motivations, your preferences, your dreams, hopes, aspiration, you name it, that just gives me a lot of power in trying to figure out, here's where I might be able to push them. Here's some of the ways in which I can communicate. So once we understand who's on the other side and what they might care about, and that just makes it a lot more likely that we can communicate in a way that's persuasive and compelling.
Now, in the context of data that can take so many different shapes, that could mean that I use it to sell you more products, because I understand why you would want to use a certain product, right? So one of the studies that we did was with a beauty retailer, and it was like a very generic campaign that tried to get women to click on an ad on Facebook, have them go to the website and buy something. And just by distinguishing between women who were more extroverted and maybe interested in makeup to have an amazing night out with friends, be the center of attention, be the queen of the dance floor, they might respond more to these types of messaging versus women who are more introverted, who probably use these beauty products to make the most of the me time they that they have at home, and it's much more internally focused, and they're probably going to respond to different types of messaging. And in this collaboration with a beauty retailer, exactly what we found. So same product, we're just talking about it in different ways, and that got people to click more often and also eventually purchase more often.
BH: How else are the algorithms being used to shape behavior outside of commercial purposes.
SM: The big question right from the beginning was, what else can I use those algorithms and like the mechanism of psychological targeting for that's not just selling products, right, and to do in a way, the most extreme opposite example is, well, could I also use the ability to peek into your psychology and change your behavior to not get you to spend more, but actually accomplish the opposite and get people to save more? So one of the projects that we've done with my team is essentially teaming up with a with a FinTech company that's trying to help low income individuals save money for a rainy day, which is one of these behaviors that is just so hard for the brain to do because you have to make a sacrifice in the here and now for like this potential benefit in the future. So what we did with a with a FinTech company is very similar to what we did with a beauty retailer. We predicted people's personality, and then we came up with messaging that was trying to highlight different reasons for why people might want to save. So if I have someone who is very agreeable, that's one of the personality traits that characterizes people who care a lot about their social relationships. For them, it's not necessarily just putting an extra dollar in the bank account, if I can tell them, look, the reason for why it might be helpful for you to save is because it allows you to protect the people that you love in the here, now and also in the future, that's probably going to go much further in convincing them to actually put that dollar in the bank account. And again, similar to what we saw with the beauty retailer, where we got women to, in this case, spend more money, we also, in the context of the FinTech company, managed to help people save more.
BH: I want to ask you, what the impacts are for business leaders and how they can balance the risks and opportunities of these technologies?
SM: For me, it's one of the most important questions. Because I do think that we're going to need regulation to make some of these things easier for businesses and consumers, butt here's all of these like new technologies that I think are incredibly interesting for business leaders who care both about providing this exceptional, personalized service and convenience that consumers are demanding and craving, but also in a way that protects basic privacy rights and some level of self-determination, agency. And the one technology that I'm thinking about in particular is something that's called federated learning, which is a way in which you can essentially generate intelligence from data without having to collect the data in a central server yourself. So Siri, for example. The way that Apple trains its speech recognition algorithm is not by grabbing all of your speech data, sending it to a central server of Apple, where they then train the models and make them better over time. What they do is they make use of the fact that your smartphone is an incredibly powerful computer. So the computing power of your smartphone is so many times higher than the computing power of the one the computers that we used to launch rockets into space with just a few decades ago. So what Apple can do is it can essentially can take the predictive model, the speech recognition model, send it to your phone, and now it locally on your phone, takes in all of voice data, it learns to better understand you, learns to better respond to you. Your data never has to leave the phone. What Apple then gets back is essentially an updated version of the model, an updated version of the intelligence. So it learns something from you, sends it back, so everybody else also benefits, but in a way, it gives you again, the personalization and convenience, but the data itself never leaves itself “Safe Harbor”. And for me, that's a total game changer.
If you think about data, once you've collected it in a central place, it's your responsibility to protect it, and there's going to be a lot of entities out there trying to steal it. So if we just look at the cost of security breaches, and it's purely financial costs, not even talking about reputational costs, you're much better off not sitting on that pile of gold if you can provide exactly the same service and maybe even make it part of your value proposition to consumers, the way that Apple is doing. So if you look at the ads that Apple is running, it's all about how they protect the privacy, and you can talk about them and the incentives behind that, but essentially, I think for companies, there's a good argument to be made that you're better off providing the service with the intelligence without collecting the data itself.
BH: That’s such an interesting example, Sandra, what about the risks for organizations using the information for, say, nefarious purposes?
SM: I think for anyone who uses personal data for nefarious purposes, I think there's always going to be a reputational risk. Because it's essentially once that comes out, even though I would argue that consumer memory is really fickle and moves on way too quickly, I think you might be losing trust really quickly. But almost the perhaps more interesting question is even for the companies who use it in more benevolent ways, there's still a risk of consumers having this feeling that you're they're just being exploited and manipulated behind their backs. Take the savings example, right. Even if you're trying to help me accomplish something that I'm hoping to accomplish myself, just me finding out that you're doing this behind my back might feel like it's a breach of trust. So whenever companies can, I always recommend making it both transparent and like a two way conversation. So coming back to this question of accuracy. Those models are never perfect, and you're going to get it wrong. And that's frustrating for both your customers, because they're not going to see what they want to see, but it's also costly for you, because you're not optimizing for people's actual preferences. So if you can make it this two way conversation, say, “Look, what we're trying to accomplish is improve your experience, but trying to help you achieve goal X, and here's how we're doing it. Here's the data that we're collecting, here's how we're using it, here's some of the predictions that we've made. Now you interact with those predictions as much as you want. Tell us if we're getting it wrong. Tell us if there's something that you don't want us to use in the recommendations that we make.” That's not just good for consumers, because they get a lot more control. That's also very helpful for businesses, because it allows them to improve accuracy, while at the same time, also increasing trust.
BH: So much of this happens, though, behind the scenes or in the in the digital systems, and it's just not visible to consumers. Do you think that there needs to be more awareness and more focus by all of us, and maybe even especially younger generations, in terms of understanding how their data is being used, so that they can have those pressures and perspectives to share with companies.
SM: I agree that a lot of this currently happens behind the scenes, but I also do think that you actually see companies trying to differentiate themselves. So if I look at all of the ads that I get on Instagram, a lot of them are, essentially ads for here's how to dress more nicely. I'm like, thank you so much. But what they're trying to do is essentially, a three-minute survey that says we want to try to understand what you're looking for. So it's very explicit in terms of both the data that they're collecting and the goal that they have in mind and the goal that they set for myself. So I do think that there's more and more companies that are trying to move away from this model, where it just happens behind the scenes. In terms of awareness, I have become, quite like a lot more pessimistic, I think, over the years when it comes to transparency education, not because I don't think it's necessary. I think it's absolutely necessary.
The more that you know about how the game is being played, the better you can protect yourself, the problem is that technology moves so fast, right? So I think about this pretty much 24/7 and I have a hard time keeping up. And even if you manage to keep up, if you really wanted to properly protect your privacy, that would be a full-time job. If you really wanted carefully read all of the terms and conditions, think through all of the permissions that you kind of grant when you download an app, that would mean like a 24/7 job. You could say goodbye to all of the fun family dinners and movie outings, because managing your data would be all that you'd be doing.
So I do think that we need transparency and control, but I don't think we can just put all of the burden on consumers. I think that's when regulation comes in, and that's where hopefully also business leaders just step up and say, how do we make it easier for people to do the right thing?
BH: You're involved in this world every single day and researching it. So how do you manage your own data profile and your own privacy, and are there any practical tips there that you can share with our audience?
SM: I mean, I think my own life is one of the reasons for why I've become a lot more pessimistic. I think of the dark side if you want, almost every day when it comes to privacy, my ability to make my own choices, and I can't keep up. Most of the time I try to do the cookies, but there's so many times where I just don't have the energy and I just don't have the patience. So the one kind of space where I might be a little bit more mindful than maybe other people is really the phone. I think so many people are focused on social media, and there's always when I when I give talks, there's someone who's very excited and that they're not using social media, and they feel very protected. And then you tell them, “Look, you have a credit card that you swipe every day. You have your phone.” And on the phone, like most of us, again, we download an app, and we mindlessly accept all of these permissions, tapping into your microphone, tapping into your photo gallery, recording your GPS records. And if you think about the offline equivalent of that. This is like someone looking over your shoulder 24/7. Someone opening your mail, reading your mail. Someone interfering with the connections that you have and peeking into your photo albums from the last 10 years. It's incredibly sensitive, so almost more than social media, which is something that you can like somewhat more easily manage yourself. So the one thing that I'm probably a little bit more careful with is really the phone.
BH: That's great advice. And as we come to the end of our time here, I'm wondering if you have advice for us, not just as consumers and users of these platforms and phones and data systems, but for people who are leaders and are in a position of responsibility to oversee the privacy and data collection practices of their organizations, and people who aspire to those roles, what is one thing that you would like them to keep in mind?
SM: It sounds trite, but it's essentially, don't stop at what's legal. I think there's so much talk of like, Oh, do we comply with regulations? What you really want to aim for is, here's what is ethical. And it's really hard to sometimes anticipate this, especially when you're excited about the launch of a new product. So everybody's excited, because people have been working on this for a long time, so down in the weeds developing it, that you see the benefits for consumers, but it's not always immediately obvious for people so involved, like, here's the potential risk. So coming back to Apple there's a thought experiment that they go through, which I think is just such a nice way of anticipating challenges when it comes to the collection and use of data and they call it the “Evil Steve test.” Whenever they have a project and they collect user data that's potentially sensitive, they ask themselves, what would happen if tomorrow we got a new CEO with completely different values, completely different way of thinking about the business, not trying to help users have a better experience, but really try and hurt them, both at the individual level and potentially at a society level. Would we still feel comfortable collecting all of the data that we're collecting right now? And if the answer is no, then you might want to go back to the drawing board and see if you can do better. And I think that's it's a nice way of overcoming this optimism bias that we have while we're working on these products, and we know very clearly that it's going to help people have a better experience in the here and now, not necessarily anticipating what might be the challenges of the data being around tomorrow, because data is going to be permanent. But the leadership using the data might not be.
BH: That's a great test and example, I think, for people to keep in mind, and it really surfaces that behind all this technology, there's an ethical dimension that we have to understand. Thank you, Sandra, for sharing all these great insights and perspectives on this technology that we live with every day and probably need to understand a little bit more deeply. We really appreciate you being here with us.
SM: Well, thank you for the conversation.
BH: Our guest has been Sandra Matz, and her new book, again, Is Mindmasters: The Data-Driven Science of Predicting and Changing Human Behavior. This has been Visiting Experts, a Rotman School podcast for lifelong learners, exploring transformative ideas about business and society with the influential scholars, thinkers and leaders featured in our acclaimed speaker series to find out about upcoming speakers and events visiting us here at Canada's leading business school, please visit rotman.utoronto.ca/eventsthis episode was produced by Megan Haynes, recorded by Dan Mazotta, and edited by Damian Kearns. For more innovative thinking, head over to the Rotman Insights hub and please subscribe to this podcast on Spotify, Apple, YouTube, Amazon, or wherever you get your podcasts. Thanks for tuning in.
SM: Whoever collects the least data on you…