Better Data, Better Date?

Introduction

0:01Welcome to the Harvard Data Science Review Podcast. I'm Liberty Vittert-Carpido, and I'm joined by Miguel Pérez, standing in for our Editor-in-Chief, Shaoli Meng, while he is out.

0:13Today, we're diving into the most personal applications of data science, love. We've all heard the statistics, millions of swipes, thousands of matches, complex algorithms promising to find your perfect partner. But what does the data actually tell us about human connection, and can an algorithm ever truly understand chemistry? To explore these questions, we have two fascinating guests who approach matchmaking from very different angles. Amy Anderson is a professional matchmaker who has spent years using intuition, personal interviews, and human insight to connect high-profile clients with compatible partners.

0:50Adam Cohen-Islati is a 17-year veteran of the dating industry and CEO of Three Day Rule. Together, they represent a fascinating tension in modern dating, the art of human connection versus the science of algorithm matching. Whether you're happily coupled, actively dating, or just curious about how technology is reshaping romance, this conversation will give you a whole new perspective on the data behind our love lives. Let's get started.

Matchmaking Experts

1:20Amy and Adam, you've both, I mean, you've spent years helping people find love, whether it's through high-touch matchmaking. Adam, I know you used to be around algorithms. Now you have more sort of a mixture. You can explain it to us. But, you know, what's the most surprising thing you've all learned about what people might actually want versus what they say they want in a partner? Well, it's interesting because I think so many people come in with this incredibly long checklist of must-have requirements.

1:57And for me, I've been doing this now about 23 years. A lot of that, I look to that information as noise. And for me, so much of this is really based on kind of the values of what somebody's looking for, based on a really significant intake process of getting to know them. And sometimes people don't even realize these kind of core values up front. And it's really through our extensive conversation, typically in person,

2:30that I'm able to extract kind of these key pieces of data that I think somebody needs. So the human versus the machine is able to really figure out the essence of what's going to be the core DNA, that's what I would call it, that is going to hold two people together for the long haul. Okay. It's interesting because I started my journey of bringing people together as a grad student at Harvard in 2007. And what I realized very quickly when I was 22 years old was that dating is a skill.

3:07And the skill of dating is not taught in a high school, is not taught in a college, and really should be. Because we're learning how to date through the media. We're learning to date through trial and error on apps that really are swiping us to nowhere. And we may learn how to date from our family or friends, which also can be good examples or not so great examples. And so how do we know what we really want in a relationship unless we see it? If you see in your family, your grandparents have been married for 50 years or your parents, that's wonderful.

3:40Most people don't have that. 60% of people don't have a grandparent that's been married for a long time or a parent that's been married for a long time. So how do you know exactly what you're looking for? The truth of the matter is you do not. And I think that, you know, when you end up not knowing what you're looking for, in your mind, you're thinking, well, my mother says I should marry a Jewish person because I'm Jewish or an Italian person. Keep the religion, keep the faith, keep the traditions and the values. Those things can be important for some people, but it's not enough for love. And so I think what happens when you go to a matchmaking business,

4:14and three-day rule is probably the third or fourth largest matchmaking company in the country. And we've done this at scale for 15 years and we have over 20,000 successful matches with humans, right? Forget about all the technology that we're building, but just with humans. So to Amy's point, we have these data points. We know what exactly is a good relationship. And we try and educate our clients on how to date successfully, how to date better, how to show up better.

4:44What are we looking for? And so in the case of three-day rule, we have a three-day rule approach to dating, which is what you should do the day before the date, the day during the date, and the day after the date in order to achieve dating success. And if you follow this three-day process, your chances of finding love, your chances of that date going well, which means you both want to see each other again, go way up. And the reason I bring this up is because for many people who are single, they may say, well, it's the other person.

5:15I'm not going to settle. And I hear this, especially with young people. I'm not going to settle. You shouldn't settle. But I think knowing what you're looking for and not looking for red flags is important. Young people have grown up on dating apps and dating apps have told us that there's a million fish in the sea. There's literally an app called Plenty of Fish. And what that does to humanity is it minimizes a human being into an object. And as soon as you spot something that could be a red flag, you run the other way because there's a hundred million other people to date.

5:49Matchmaking turns that whole component on its head and says, the essence of dating, the essence of successful relationships is people. When you give people a chance, when you get to connect with people, when you throw out your inhibitions and you really give that person a minute of your time, learn to listen. Dating success can be there for everybody. And so with technology, we're able to take the amazing work that Amy does and companies like Three Day Rule does and make it affordable to so many people

6:20who can't afford $5,000, $10,000, or $50,000 with a matchmaker, but take the same essence and democratize that. Because when you're sick of those apps, where do you go? It kind of feels like this necessary evil. And you've been swiping for five years and you're still single. And if you don't change your method, you're not going to change the outcome. And so we're very much set on changing the method to change the outcome. And I'll also kind of extend to that. So my motto at Lynx is do what your grandparents did. Chances are when your grandpa was courting your grandma, he kept things really simple.

6:53There was a clear signal that he was interested and he asked her out. Maybe he gave her a rose and she said, I would love to see you again. She leaned into that opportunity. There wasn't the paradox of choice, the millions of different suitors around her. There was maybe two guys and she expressed interest in the very strong signal pointing towards her. And therefore, they made things very simple, not complicating dating, right? And so I think when we remove a lot of the chaos from around us and we distill things down to what I really call the basics

7:31and lean in with intentionality for how we date and think about finding an appropriate match, that's when my clients see results. It's like stop playing games, pay attention to that person who is expressly interested in getting to know you and really see if there's some merit there. A huge part of what I do at Lynx is also the strategy that goes into this.

8:01But that's really the human approach in what makes high-end matchmaking really special. Because I feel without that, kind of the human sharing my level of expertise, people can quickly get derailed in default into something, you know, that is really old habits and patterns of, again, trying to kind of overcomplicate something that doesn't need to be as complicated. And also, I'll just add one more. We love data here. I'm a data nut.

8:32So matchmaking has a between 70% and 80% success rate across our industry, and dating apps have a 9% success rate. So what Amy is telling you is if you want to be successful, slow it down. Take your time. When we say intentionality, everyone rolls their eyes. Why do we need to speak to 15 people on an app? You can speak to two highly qualified people and go on a date with one of them, and that's going to be a great match for you. Slow it down. Be intentional.

9:04Matchmaking is a more efficient way of dating because you're only getting people that Amy or 3DayRule have dated for you. They've been vetted. They've been screened. They've been personality checked, credit checked. We know that there is compatibility. So the baseline is 10 times higher than the baseline of you doing it on your own. Exactly. And Adam, one thing I was going to say to that is with our businesses, high-end matchmaking, you know, the way that I think about this business is really kind of from the inception becoming the ultimate force multiplier through signal compression.

9:37My clients actually tell me that they want a reduction in the number of matches, you know, as compared to kind of high-volume apps, to your point, and much more kind of highly targeted individuals that could be high-probability matches where the chances of actually finding love is going to greatly increase as compared to a bunch of lower-probability individuals where they might never, you know, find any sort of connection.

10:09So I think that that's been a really interesting discovery, especially with the rise of apps and just so many different kind of choices out there that clients come to me like absolutely fatigued from their previous set of experiences and just saying, listen, I'm not looking for 15 or 20 or 30 introductions. Amy, if you can actually give me a curated pool of four, fine, maybe five individuals that are high signal and could be the one, bring it on.

10:40And I'm like, wow, I love that. So it's been really interesting to just see that kind of evolution over the last handful of years.

Surprising Data Insights

10:48Adam, when you were at Bumble, this platform generates a ton of data. So what's one pattern or insight from that data that genuinely surprised you even after years in the industry? So when I was there at the company, I was a managing director. They owned four apps. So I ran one of them. And so while they have a lot of usage and behavioral data at the time, the biggest piece of data that they used was swipe data, not conversational data, not profile data.

11:18What's going to lead to more swipes because more swipes is more time on app. More time on app is more inventory of people. And that's potentially more opportunity to upsell you to premium. So it was a little bit shocking to me, not only there, but in other places that they actually use very little profile data at the time in order to match people. It was kind of like, well, you like sports and long walks on the beach and you're the same age in New York. So you must be a good match. There's no reason why these two people should ever be compatible, but that's really, it's

11:49very, very basic. And I think that, you know, with all these casual apps, they're using swipe data or click data, depending some of these apps are grids. So either click or swipe data to see what kind of people you're interested in, because the more they keep you coming back to the app, the more potential either for ad revenue or for you to upgrade. So on Tinder, every 10th swipe, if you're a heterosexual guy, it'd be a really good looking girl because after 10 swipes, if you don't see someone attractive, you're off the app. So the 10th person is going to be attractive to keep you swiping more.

12:21And then once you hit the 20th person, they're going to upgrade you. So they're looking at user behavior, but I don't think from the right point of view, they're not really looking to connect you long-term compatibility. And that's because the whole point of matchmaking is we want to get you to a relationship and off of our books, there's millions and millions of singles in the world. I don't need more singles. I need more people in relationships. Amy, you and I have talked a lot about the amount of data that you use. And a lot of this data is not digital, but it's a lot of data, very important data.

12:53I'll throw the question to you. From all your years of experience, 33 years, what are some of those pieces of data that have surprised you and are the most predictive of good matches? Mm-hmm. Well, I think that a lot of people say they want chemistry, right? But at the end of the day, I think one of the surprising kind of truths out of this is they really want emotional safety and somebody who's kind of accountable and emotional maturity.

13:25So I'd say those are some of the key things that I've seen over the years that people really want, right? They can say they want something, but at the end of the day, there's far more important things that they kind of need for a substantial and a long-term relationship. You know, it's funny. I feel like for the first time on the Harvard Data Science Review podcast, we found a situation where the algorithms are not the way to go and that it's truly this sort of personalized

13:58human touch. But at the same time, really what both of you all work with, what matchmaking really is, is it is data. You know, I called it human intuition or, you know, whatever you want at the beginning, but it's you all have collected years and years and careers of data on people. So I think, you know, in the same sense, it's, you know, so much of what you all have talked about is sort of like people don't know what they want and then you all through matchmaking help them. But has there been a situation where the, you know, sort of almost the opposite, like where user behavior or client behavior is a better word than user.

14:30I'm still going to the algorithm word, um, where client behavior has changed your data. So just, just to kind of be clear, we are a three day rules, a very data focused company. So we actually took 15 years of our client data. We know matchmakers have something special that dating apps don't have. And what it is, is called post date feedback. I'm sure you guys remember millionaire matchmaker with Patty Stanger, right? So on the show, after the date occurred, she called both of them and she said, I want to

15:03know what happened, why it worked, why it didn't work. Give me as much information as I can so that I can iterate on this. So we do the same thing. And so does Amy, I'm sure at our company, we give, we actually give you a survey. We actually, because we want to ingest it into our own system. We want to find out based on two user profiles of demographic, psychographic, and all these full on details, when the data occurred, was it a yes, yes, they both want to see each other again, a yes, no, or a no, no. So we built our own three day rule matchmaking app based on 15 years of our own data.

15:40So we know exactly why two things worked or didn't work based upon your, your specific person, which could be, I'm a Jewish woman from Boston and I have these traits and I like these, whatever, this is what my career trajectory is going to be. We can find you your person. Now we still have to have that person in our network to match you with, but we know what kind of person will be compatible to you because we've seen it 20,000 other times. So that is our app that we built. And that is at scale for everyone across the country, every single. And then in terms of within our own business, we make changes all the time because what

16:15Amy's referencing is matchmaking is magical. There's an X factor that we cannot tell you about. We can give you all the data points you want and maybe 80% of the time it'll work, but for 20% of the time, it won't work because you're a unique individual and you're not a clone and you're not made from someone else. You're unique and your experiences are unique. So we don't exactly always know what your behavior will be in a date or what your attitude will be or how you're feeling that day. There's so many factors that can influence how you perceive someone else.

16:46So we try to take as much into consideration as possible, but if you're having a bad day at work, probably don't go on a date, go postpone it because you don't want to waste that person. There's so many external factors that can influence whether a date was successful or not. Amy, you work with clients that are overwhelmed by choice. And many of them are very, very successful in their careers and professions. And Adam, this is true for many of your clients as well. What patterns do you see in these endless options that the apps provide that are having

17:19your clients go to you guys? And what are they hoping to get that they just have not been able to get out there? One big thing at Lynx is there's like a method to my madness, which is the amount of homework that I have candidates complete and clients complete. And there's a reason behind that. It's going to weed out individuals who are not aligned with our values, who do not see the

17:50value in this, who are not willing to kind of put that initial time into it because maybe they're lazy. They're just whatever, whatever their rationale is. And so from the inception, we really align value sets very carefully. So when we do match, the client knows whoever they're meeting has gone through a similar, if not the same system. And right there, that just provides a lot of relief and kind of mental, I think mental

18:23relief, like that anxiety is really dissipated or the kind of questioning, is this person kind of sharing the same values of what I'm looking for? Like from the start, everybody knows that anybody who they've been matched to, they've gone through the same system. And so it just alleviates a lot of kind of concern and worry. So to, again, kind of protect well-being, mental well-being, integrity, heart, and just

18:55hopefully have things already be off to a very strong start. So Adam, Amy, is AI and using AI in the process of getting to know people and potentially writing prompts and writing responses and maybe just like doing things for one another and posturing this towards the other person. Could this be the new catfishing? How does one think about this? What's ethical? What recommendations do you guys have? I mean, I think it's an enhancement.

19:27It's not a replacement. I think it's obvious when somebody's used that, right, for responses and whatnot. I mean, you can kind of sense that or just instinctively know that somebody wrote a response had AI do that for you. So I think it's a nice kind of augmentation, but it's not everything. I have some clients that lean very heavily on AI, like ChatGBT, to help them just kind of parse through their love lives. You know, I provide my perspective, but they're like, well, ChatGBT said this or that.

20:03So it's, I think, a nice add-on. But at the same time, I think it does become a very slippery slope and potentially dangerous if somebody is relying so heavily on it for certain answers or validations or, you know, decision-making, people need to be able to rely on themselves. And even if they feel that they don't have the answer themselves, like, ask your friends,

20:34like your trusted inner circle who knows you well versus the machine. I think, again, augmentation, not a replacement, but use with caution. I think it really depends. If you're referring to how AI is being used with dating apps, I do think that the current crop of dating apps, because verification is not a requirement, you don't have to use Apple Connect to create a profile. Any random person can. If a bot can create a profile or create onboard and create a profile and then a bot could write

21:07everything in your profile and match you, basically swipe for you, which can already be done, that is going to be a huge, all these dating apps are going to be full of bots. If you're doing something like 3-Day Rule is doing, and we, you have to use an Apple, you have to have an Apple Connect account to be able to use this app. You have to have a verified photo to use the app. You have to have a completed profile to use the app. It's a very different way of using AI. We're using AI to augment how you show up. What's happening on a lot of these apps for the regular person, they don't, they don't

21:39show up correctly. They don't know how to show up. So oftentimes people show up very generically and their difference is the photo. And so on an app like 3-Day Rule, we have voice AI and we have a 10 minute voice conversation with you to not only get who you are, the essence of who you are, but in our app, we look for your tonality and we actually have a lie detector built into our app. So if your tonality changes when you're giving us a response to a question, we'll come back and we'll say, hey, that tonality changed by over 50%.

22:10We're going to re-ask that same question in three different ways. A good matchmaker also knows when their client's feeling anxious about answering a question, she'll put a pause or he'll put a pause in it and come back to that area, that topic later on. Because if we don't get a good read on who you are, we cannot match you. If you tell us what you think we want to know, we're not going to match you well. So it is very important and our app is built the same way that a matchmaker would interact with someone. So in that way, I think AI can be very additive, but just like AI is for any person working at

22:40a business, if you're not learning and you're just using it to do your job, you are not going to have a job for very long. It will replace you. So be learning while you're using AI. If you're using AI to help to write a profile, learn how to write a good profile, learn how to give a good response. Then you're benefiting from it becomes an educational tool versus a tool that leads you to nowhere.

Future of Dating and AI

23:02All right.

Future of Dating and AI

23:02So I have my, my, we always sort of have this magic wand question at the end. Um, and it's, it's asking you all to wave your magic wand. Um, so if you could wave your magic wand and create an AI tool that would help those who are struggling with dating and, and I'm going to add this in there that would also help with the future of dating, what would it be? Well, I'll start. I mean, or is there one?

23:34You're like none. Just skip the way back. When I really think about the intricacies of like dating and feedback and all sorts of things, I think that there's a lot of people who would ask the AI tool certain questions that they would not ask the human, right? Because of fear of, you know, my judgment or something that, you know, it would be deeply revealing that they could be embarrassed about or whatever it is.

24:07So, um, when I think about kind of the future and waving that magic wand, it is some sort of tool like that. Um, giving individuals, whether it's the masses who are not my clients or my clients, kind of that opportunity to gain extra insights and talk with this, you know, talk with the machine, um, about certain things they would not necessarily reveal to me, um, which I think is really cool and very exciting. And the way I'll answer this question is a shameless self-promotion.

24:39We've already built this. So 3-Day Rule built the first AI powered matchmaking app built by matchmakers and 15 years of their own data. So we created a date coach on our app, which is separate from our matchmaker who will walk you through what should you do on a date? How should you respond to a question? Is sex on the first date? Okay. If you want a long-term relationship, whatever makes you feel anxious or unsure on demand, you get all of your questions answered instantly in the way we would answer those for white glove

25:12clients. And it's cost you $0. It's free. Matchmaking is literally the future. Now, how that comes to play with AI, with non-AI or how much AI you use, that's debatable, but the future of this industry is matchmaking. And the reason for it is actually does work. I love it. This has been super fascinating and a really interesting aspect of, of, of data. So thank you both. You got it. Thank you so much for having us here.

25:44I mean, man, that was a fascinating conversation. And the, the tension between these apps and matchmaking and human intuition or what we now sort of think of as data. You know, it's clearly not just about data. It's really about whether the most important parts of human connection can be quantified at all. But I really want to thank Amy Anderson and Adam Cohen-Islati for sharing their insights with us today. You know, if this episode has you thinking differently about how algorithms shape our choices and dating

26:16and beyond, that's exactly what we're here for. So if you enjoyed this episode, please subscribe to the Harvard Data Science Review podcast wherever you listen and consider leaving us a review. It helps other curious listeners find us. You can find show notes, transcripts, and links to our guest work at our website. And if you're interested in diving deeper into how data science interacts with human behavior, check out our episodes on everything from predictive policing to personalized medicine. So until next time, remember where you're swiping or being set up, the data can guide you.

26:48But the chemistry, that's still all you. Harvard Data Science Review, everything data science, data science for everyone.