Why 95% of AI Projects Fail and How To Beat the Odds

Speaker A: If we think that the skill set that brought us here is the one that will keep us relevant in the workplace, then we'll have problems. And, uh, education is not helping. Corporate trainings are not really helping. And so it becomes a very individual decision. And, uh, it really depends on the sense of urgency that each one has.

Speaker B: Welcome to Modern Business Operations, where we talk with leaders about how ops is adapting to our modern world.

Speaker C: Hello, everyone. Welcome to another episode of Modern Business Operations. My name is Sagi. I'm the CEO and founder of Tonkin. And today I have the great pleasure of hosting Andrea. Hey, man, how you doing?

Speaker A: All great, Sagi. Glad to be here.

Speaker C: Well, thank you for, thank you for joining me today. There's a lot of things I want to jump into. A quick, quick, uh, overview of your kind of recent career. You have, you're the author of an upcoming book. Between you and I want to talk about that. Very, very, very timely. Uh, but you're also used to be the head of Tinder Latin America and the chief Digital Offer Officer of l' Oreal Brazil. And you write a lot of things for a lot of different, you know, known magazines and you do a lot of keynotes. And so there's a lot of things I want to, you know, jump into. But maybe we start, you know, over overview who you are, how did you get to, to do all this stuff and maybe kind of like a short, short, uh, timeline.

Speaker A: Sure, Sagi, it will be a pleasure. I mean, my name says it all. I'm Italian originally, but, uh, I've spent, um, 10 years in Brazil where I actually kicked off my professional career. I'm an economist by education, but I've never worked with that. I actually pivoted to, of course, business, but especially, uh, technology. I was the first employee of Tinder, the dating app that some of the listeners know about. Or maybe we can share some tips and tricks now, maybe towards the end of the episode. Sagi.

Speaker C: Right.

Speaker A: We'll see whether that's interesting to them. But jokes aside, I was the first employee of, uh, Tinder in Latin America and then scaled the operation, uh, for five years. We took it to, you know, become the top grossing app on the App Store in Brazil, Argentina and Mexico and a number of other smaller countries. But I was also the Chief Digital Officer at l' Oreal Brazil. And in parallel to that, I started doing some keynote speaking, which eventually became my main activity over the last five or six years. And in, uh, parallel to that, again, I write for a number of publications such as the MIT Technology Review and uh, the Wired magazine. And um, again this book is coming out in November called Between youn and AI, uh, which is about the skill set for professionals and leaders to thrive in the age of intelligent technology. And we'll talk further about that. But uh, yeah, on a side note, I'm uh, black belt in Brazil Jiu Jitsu. Life is not only about work. I'm based out of Miami and uh, here we are. It's such a pleasure being here, Sagi.

Speaker C: Awesome, thank you for that. Hey, do you introduce them? M. That's actually interesting. I'm wondering how much that impacts your uh, philosophy of life. But maybe let's double click, you know, into, into the book. I interview a lot of folks from all different areas in, in this podcast and obviously in the last year and a half, every conversation very quickly ends up talking about AI. Uh, even if in the pre show we say we're not going to talk about AI this time, it kind of ends up talking about that too. And I think there's a, you know, kind of like divides between, you know, excitement and fear and everyone has a little bit of both. Different percentages maybe with that know, what is this book about? You know, what got you to write it and what, what are we cover? What are you covering it?

Speaker A: So basically, Sagi, as much as uh, you've said, uh, as soon as AI started to basically become of course massively used, especially with the launch of ChatGPT, its first version around 2022, I started reading a lot about it. And most of the books, actually all the books that I was reading were books focused on the technological aspect of it. Of course, many also on the business aspect of it. But I wasn't really finding any book that was talking about the human aspect of it. Of course some would be touching upon the impact, you know, AI has on the job market and on uh, the future of work. But uh, I was really intrigued by the impact it would have on our human skill set and especially which new skill set would it uh, require for, you know, from successful professionals. And uh, because of the fact that again after my career in tech I started to dive deep, uh, maybe the fact that, you know, I'm an economist by education really helped on that because actually my field of expertise is behavioral economics, which uh, Daniel Kahneman, Dan Ariely have been popularizing over the last years, but really focuses on the behavioral aspect and you know, human skillset. I merged the two things and I really uh, started doing research and uh, uh, of course writing on this, uh, unique intersection between AI and so mapping out what AI does better than humans, but really in detail, uh, getting to see for example whether humans or AI is better at adaptability, which is something we can further talk about. And uh, where humans are much better than AI, of course, also focusing on what AI does best, which is ah, substituting and learning on the very repetitive tasks and the familiar tasks where you have uh, uh, basically a uh, measurable outcome because then you can feed the data into the AI and AI learns which actually take up around 40% to 50 or maybe even more percent of our daily tasks. And so that's where a lot of people are scared because they'll be like, okay, then I'm going to be substituted. No, a chunk of the tasks that we perform are going to be substituted. The problem is though, if we don't compensate that with uh, a uh, whole new set of tasks, then we, we will become more and more worthless in the uh, job place. So I focus around three big pillars of transformation, what I call the cognitive, behavioral and emotional. And within them I lay out nine new skills that become more and more important in the age of uh, artificial intelligence.

Speaker C: So that's again very timely. Let's maybe half a step back. Do you think it's, aside from writing these books, we obviously believe in this, the importance of this, but how much do you think the hype versus not hype. Do you think this is really what it said to be? Do you think we're over, you know, we're overexcited? How do you rate the reality versus the uh, potential?

Speaker A: So if we look at the recent stats and uh, you know, we're recording this now, mid September 2025 and uh, truth is the latest numbers are not really reassuring on the actual impact. So pointing to sort of an AI bubble or hype, which doesn't mean the technology is not transformative, but which might mean that, you know, valuations out there are super high when it comes to the startups and the investments. Also the corporate projects seemingly According to an MIT Media Lab's recent report, 95% of you know, AI projects, uh, actually fail. But the interesting part of it is that I even think that maybe this overhype is actually good because it does a number of things. First of all it points to the things that we're not doing right through the adoption. And again when we look at the reasons for failures of many of these projects are first of all because the companies that use the tech tools do not have the data ready. And so it's bringing uh, a uh, Spotlight on the data readiness AI, data readiness of companies. Second, which is a topic you guys are very big experts on, is the operations and processes are not in place. And the third big aspect is, again getting back to the topic of my work, is the human aspect of it. There's not enough training, you know, out there, or at least the training is too technical and too little behavioral. And uh, the consequence of that is that many of these projects fail when it comes to the measurable impact of them. And so, you know, there's this theory that, it's called the bullshit job theory. I'm sorry for that. I don't know whether I can say that, but I know I can. But it says that basically and it is actually corroborated by a more academic theory which is uh, by Darren Acemoglu, the simple microeconomics of AI, which says that basically AI came to make more productive jobs and tasks that were already very unproductive and therefore the real productivity gain of AI is not that real because those tasks and jobs were very unproductive in the first place. So I think, yes, getting back to your question and wrapping it up, there's a lot of excitement, maybe also too much, which points to a potential deflation of a bubble. But as much as that happened with the Internet, uh, you know, in the 2000s, I think that might be actually good for the technology.

Speaker C: Yeah, I like, I like it to kind of finish with the Internet because I was about to go there too. Right. There's a clear distinction between whether the, the market is over, fuming with, you know, with excitement and whether or not this technology wave is going to be transformational. Like the dot com boom was one of the probably biggest one yet. You know, 25 years later to think that, or even 10, 10 years after that. Right. But like know now 25, we have the, the ability to look back 25 years and everything is different. Like there's not even one thing, one aspect of our life that is not different because of, of, of the Internet. You know, the cars are using, you know, Internet. You uh, know, so it's like it's everything is using, is using Internet. And so I, I, I, I like that comparison a lot. Let me double click though on that last point and kind of like further into the area of work you kind of COVID in the book. I do think that similarly to that Internet revolution, whether you believe the AI revolution is happening right now will happen only in 10 years or anywhere in between. It's very clear that some things will change and it does Feel like really stood out for me. The nine skills concept that there's actual the humans needs to now really look in inside of where we need to be better. So let's expand on this a little bit. What are the key findings there? Obviously as an introduction to the book and the concept from a high level, that's perfect.

Speaker A: I mean basically whenever we look at the comparison between the way that humans uh, work, namely on a cognitive aspect, on the task execution aspect and on the sort of social aspect, there's many differences between uh, the way that AI works with respect to our human brain and the feelings also that we have. And so whenever we map out, first of all on a cognitive level, AI has a very high degree of substitution when it comes to human abilities. The thing though here is that what becomes more and more important on a cognitive aspect is the ability to ask questions. And that gets a little bit into the prompting uh, space which again oftentimes the outcome of an AI tool is as good as the question that you ask it. And so that's the whole prompting uh, ability which is one of the skills on a cognitive level that I talk about. The second one is what I call data sense making. It is how well do we understand the data that we have at hand and how critically are we able to evaluate an AI uh output whether it's a result of an hallucination or not, whether it's biased or not, whether it is a consequence of a generalization problem and so on. There's this great quote by Carnegie Mellon professor Potion Law that says that we shouldn't just teach kids to solve their homeworks but actually to you know, grade the homework, which means, you know, we need to be more critical. And now part of the big problem that we see within companies and organizations is that we have a uh, workforce that copies and pastes, you know, gen AI tools and you know, just without any level of critical thinking and ends uh, up spreading hallucination and biases and so on. And that's a great risk. And the third cognitive skill uh, that I talk about is actually reperception is the ability to update our thinking in face of new information. And reperception is more and more important because AI is accelerating the rate at which things that were up until a couple of years ago, impossible, possible now. And so our mind though is being, because of heuristics, uh, program the two actually again uh, build up beliefs because it makes our decision making faster and more efficient. But now we need to also not or just to be able to make decisions but also to give up our past decisions, and that's reperception. And then maybe, you know, quickly, on the other two pillars, on the HD execution, what I call the behavioral pillar, I talk about adaptability. I talk about augmentation, which is very much related also to the operation aspect and to the work that you guys do at Tonkin. We can further talk about maybe augmentation and uh, antifragility, which is the ability to learn through mistakes. And last but not least, on the emotional pillar, I talk about empathy, trust and agency. The ability to take responsibility, this last one, to take responsibility for AIs, uh, uh, outcomes, which again, especially in a world where AI is not deemed responsible or neither technically nor legally nor morally, we humans have to take responsibility for the AI tools we use, develop and uh, adopt. And so I think it's, you know, a, ah, very quick summary of, uh, what are some of these skills and of course we can touch upon some of those more in depth.

Speaker C: Well, I agree with every single word you said. So already. Sounds like a wonderful book and I'm definitely planning to go deeper. I have a couple questions. As you were going through it, we can talk a lot about the practicality of bringing this now, but there's something you mentioned that actually is maybe further out, but actually very intriguing to me. You talked about critical thinking, you talked about the ability to grade, and even the example you gave with the quote about the homework and the kids. You know, whether you argue that the 95% failure rate that the MIT study found is going to be different next year or not, we can talk about that. What I think is very critical. I have two young kids, you know, below five. There's no way that, you know, their future is going to be. The future when they get to work is going to be similar to what it is right now in many ways. And so how far. And the reason I was laughing to myself is you mentioned critical thinking and grading homework. It feels so archaic the way that schools are, are being done now and people talked about for a long time, but now it becomes even ridiculous a little bit thinking about what my kids are studying in school right now versus what they actually need to study. Do you think, what do you think in that? On that?

Speaker A: I mean, I think, you know, like I'm, uh, an expecting father, so I don't know in practice you have two kids, so I will know soon, uh, by December when I'll have my first, uh, basically.

Speaker C: Congratulations, by the way.

Speaker A: Thank you so much, Sa. And. But I totally agree with you because to be honest, the classroom hasn't really changed since I was a kid or, you know, some schools are testing out new things, but overall, let's make an example about this first ability that I've talked and I've, you know, studied and mapped out for the book. That again is prompting. But prompting goes much beyond just the ability to write specific and context. A, uh, rich prompt to an AI tool is the ability to actually shape better questions, even to other humans, not only to AI. Let's go maybe you know, to the classroom when I was a kid, and nowadays as well, we are actually being graded by the quality of our answers, not really by the quality of our questions. I recall when I was being a kid, I was always curious, but actually school killed my curiosity because anytime I would raise a hand in the middle of the class, my teacher would say, andrea, this is not so important now. Let's talk after this question really is not appropriate now or doesn't really make sense. And so it kills the creativity and the curiosity of kids and therefore our ability. And actually the frequency, studies show the frequency of questions. Uh, I don't know how old, uh, uh, are your kids. You've said they're both under five. So maybe the older is already asking a ton of questions, right? Oh yeah, always about the why of things and how does it work?

Speaker C: It's very early. Like since they start talking, they almost, you know, people talk about, you know, mama, baba, papa, all that stuff. The real significant word is why. Why, why, why, why.

Speaker A: Yeah, and it's funny because when we're adults, especially in the workplace, we end up never asking that question again because we think we already know the why. And we, you know, again, we think we already have the answers because we've been programmed in a way where basically the best leaders, you know, are the most experienced one with you uh, know, the more years within the company and usually they're the most knowledgeable. But in a world where AI does not only overcome our ability to have, have all the answers, but it also updates its repertoire much faster than us, then it becomes a problem because not only we're competing with AI, but we're competing with anyone else that has its knowledge democratized by the access to AI. And so, uh, I don't have a legal education, but, uh, if I know how to ask good questions, again not only to AI but also to other specialists, sometimes I'm able to actually have a more in depth understanding of maybe a legal intricacy than a generalist lawyer that has studied overall law for 10 years. And so then it puts a lot of pressure again on education. Getting back to your point, and maybe the last provocation when it comes to this is, you know, it's so hard to forecast which jobs might be, uh, you know, in more high demand in the future. So it's hard to actually shape education based on that. But maybe a hint comes from, um, a recent interview. I don't know if you've seen that, but like by Geoffrey Hinton, the godfather of AI and also Nobel Prize winner in physics, uh, in 2024. And they asked him this, like, you know, in 2035, what would make me have a job? And he said, well, study to become a plumber. And he was half joking because he said AI is not as good with spatial and physical manipulation than us humans. But, you know, besides this joke, which was not a joke because Microsoft, uh, made a report with the top 40 professions by 2035. And I think phlebotomist was number one, and nurses and so on were up there in the ranking. Even, uh, you know, jobs that, you know, they're very niche nowadays. You know, to wrap it up, there's a seed of the answer to your question in there, which is we need to carefully understand what AI does better because it's not only that it makes us compete with AI, but it democratize access to that skill to everyone else. So the skill goes down in demand. Well, because there's oversupply of it. And so we need to identify what are the new skills. And I think part of the book is related to that.

Speaker C: This episode is brought to you by Tonkin. Tonkin is the operating system for business operations, providing businesses with the building blocks to orchestrate any process with no code or change management required. Contact us@tonkin.com to learn how you can build complex processes fast. Uh, yeah, of course, again, I completely agree. I think asking the question, even if you go to the Hitchhiker Guide to the Galaxy, right? It's like the answer is 42. And what? Well, that you're asking the wrong question. And so, you know, I think that's true. I also agree with the last point you made of, uh, at the end of the day, it's not even about what AI would do or not do, is about what skills are going to be in hot demand. I want to talk about some of the other things you mentioned, because I do think they're overlooked. I really, you know, lean towards couples. One, you talked about the responsibility or accountability. I think that's a big one. I want to Expand on that. And the second one was sort of like the part where it's forgotten the word you used, but it's connections to how we work as people and how you kind of bring this as part of the process.

Speaker A: Perfect. So I use the word augmentation and maybe I'll start from that because it's very much related also to the way that we look at our own workflows. So, so they're already getting into the sort of realm of operation we need to look at. Two big things that AI does when it comes to human tasks. It can either automate them, namely substitute the need for the human intervention. And that's basically where we always look at, you know, that's usually the short answer to what can AI do? It can automate tasks, and that's for sure. But it's also not only relate to AI. Ah, this is much older than the uh, recent hype of AI. What's often overlooked, especially by non specialists, is uh, the impact it has on actually not just substituting tasks, but enhancing the quality and the scalability of human tasks. And that's augmentation. And it's very important to make this distinction because if we only think about AI substituting human tasks, that's where we will have that pessimistic outlook that AI is coming to substitute us. But whenever we understand that if we automate the repetitive tasks first, we free up time to actually focus on unsubstitutable skills, but also on the augmentation aspect of it, which is, okay, how can I use AI to enhance my human work? And I'll give the example of actually Brazilian Fintech that actually became the most valuable bank in Brazil, overcoming all the big traditional banks. It's called nubank. It's actually listed in the uh, US stock market. It's a great success story of a bank that was born digital, it still is digital, but reached a huge valuation. What they do is when in their customer care, in their, you know, operations, they actually decided not to substitute the touch points with the customer with a chatbot, because that would be automation, pure automation. They decided to empower their human call center people with a panel AI powered where as soon as someone gets in touch with them, they get, first of all the whole history of that person with the bank, projected reason of why this customer is getting in touch. So it shows up in the panel. Look, I think there's a high probability that they're getting in touch because of a glitch that we had in the app. So the person is already Prepared to the problem. And last but not least, it makes you know, suggestions of uh, what the answer should be. So it doesn't take away the need of human attendance. It actually amplifies uh, the quality of their responses but also scales them and so makes one of these human beings like much more productive. Which of course eventually points to some substitute ability because you will not need 10,000 people, but maybe you will need half of it. But at the same time it still focuses on the human as a, you know, with AI as a co pilot and not definitely as AI, just as the pure solution and you know, just to wrap it up and go into what it might, uh, how it might impact people that are responsible for operations and workflows is we have to become curators of our own workflows. Mapping out the automatable tasks, mapping out the augmentable and understanding what's the time saved that we get out of that. And uh, I think that really what will make a difference in the future job market is what we do with the time that we get back. Because if a salesperson that now automatizes a lot of the invoicing and uh, you know, proposal, uh, quotations and so on gains time back, but what will make them stand out with respect to all of the other salesperson that also gain time back is how well they use that to create ties with the customer, to think strategically, to, to ask better questions and so on. So I, I, I think it will become uh, something more and more important that we look at our job descriptions and we say how much of this is, you know, potential to automation, how much of that can be augmented? And uh, it's our responsibility actually, you know, and, and leaders responsibility within the companies.

Speaker C: Look, I think, I think you know, as a CEO I can say it, but honestly, but honestly it's true for anyone, you know, know, just observing of, of nature, the energy is not going to waste. If you were able to save time or to save effort, it always go in towards investment. Everyone, anyone that thinks that, you know, you cut cost, if you're able to cut it from 10,000 people to 5,000 people, then what the company would do is just fire 5,000 people. You just don't know how business works. If I can do it with 5,000 people and I have 10,000 people, I would want to do double the stuff versus you know, doing half the stuff. It just doesn't make sense. Right. And so, so I think a lot of people that are, that are seeing the pessimistic side of it are either are either have an agenda or they're, they don't have the experience to know kind of like what's important. But I think on a personal level, you know, other than buying your book, which uh, I, uh, do encourage everyone to do, what are the things that you'll be like? This is if there's one thing that every listener should do personally to really understand. Because I'm in the belief that this change is coming whether you like it or not. Right. It's just happened, it's going to happen. It happened before, it's going to happen again. We can argue, we can discuss, we can debate the implication and the timing of it. I don't think there is a question of whether it's coming or not. So now what can you do to your, like you said to yourself, to your employees, to your kids, you know what it is? Out of the nine, what is the one thing that if you do not master by the time that this becomes norm, you are definitely behind?

Speaker A: That's a great question. So it's like the million dollar question. And in a way it all starts from asking yourself, how do I get back to differentiating myself in the workplace in a world where again it's not only me adopting AI, but it's everyone else as well within the company and in the job market. And it gets back to that productivity paradox that I've talked about before, which is as much as in the Industrial revolution before that I might have been, you know, better because of physical strength or you know, I was good at operating machines or maybe, you know, I was uh, very resistant while building uh, railroads. Well then uh, uh, mechanical innovations came and democratized access to strength and resistance. And it actually asked for more cognitive skills as a differentiator. We now live in a world where the cognitive skills are being accommodatized again and uh, the access as much as, you know, it's a uh, comparison but of course it's a much niched one with the calculators. In the 80s, right? Calculators democratized the outcome to someone who didn't know much about mathematics could have the same result than someone with a PhD in arithmetics. And it's sort of this at massive scale. And so whenever we think at what differentiates us, we need to again look at our own workflow and to our own skill set and uh, be able to critically understand that we need to focus on the what uh, AI is not able to do and what AI is not giving access to, to everyone else. And so that's why again, Ah, it gets back to some of the jobs that are much more in demand, are going to be much more in demand in the future are jobs that only few can perform today. Again, as mentioned by, you know, Plumber, because they, not only they require a very specific knowledge, but also they require a very specific ability to apply that knowledge. And I think that's maybe getting back to some of, you know, the nine skills. I would point to the adaptability as being one, because there's something that AI is not able to do is actually adapting itself, especially to a spatial environment. There's this coffee test that was proposed by Steve Wozniak as a test to measure artificial general intelligence. Right. Actually, the, you know, AGI is where everyone's putting their money now. Meta wants to build one and so on. And Steve Woznick said, well, if an artificial intelligence is able to get into someone's house, a house it has never seen before, find a coffee maker and make a good cup of espresso, then we reached AGI and AI, uh, is still very far from being able to do that because it lacks adaptability on a spatial level. It lacks common sense reasoning, uh, it lacks perception. That's where we have to move towards as human beings. And again, really being able to understand what it cannot do, but also what it does not provide to everyone else. Because that's also something that is tricky. We think, okay, well, AI is now empowering me to do this, but it's also empowering everyone else. So that is not necessarily the solution. And so the solution is going beyond and being one step forward.

Speaker C: My personal belief is that, yeah, this is correct, but it's a moment in time, right? Like the real world, AI is also going to get improved over and over with robotics and Tesla, what Tesla is doing, and other things, I think, I think we'll see, we'll see a boom in that as well. What I really like, I think the nine that you mentioned are very practical for, especially for business when you kind of trying to think about, you know, where to apply it and kind of where what the expectation should be. I really do think your first one is really kind of where it all comes down to is like, who's like, asking the questions. At the end of the day, it doesn't even matter. You know, a lot of people ask me, like, what is an agent versus not an agent? You know, all that stuff. When it comes down to it, the mission, the goal, the question, the query, the prompt, however you want to call it, someone needs to come and say, this is What I'm looking for, and this is the scope of it, and then here are some follow, um, up kind of questions. And the way you kind of like talked about the critical thinking aspect of it, and that is a skill that it's not about the skill, it's, that is the job. You know, I mean, in many ways the calculator is great, but if you're trying to just learn how to do four plus four, like using a calculator is not, is cheating, it's not helping you with the goal of the assignment. But if you're trying to figure out whether you know a ball is gonna fall and hit the, you know, in certain angle the ground, because you're not trying to learn physics, then calculating the numbers is not going to help you there. So you're going to use a tool for it. And that is a great example of just how all of human experience is already and just will continue to become more and more of that. Are you trying to be a barista? Then you're going to make your own coffee already today, right? Like you can, you can grind your beans or you can buy them grind. There's market for both, right?

Speaker A: That's exactly the point, Segi. Uh, it's exactly that. We need to understand that basically it's our own choice. Uh, it gets back to how much are we willing to understand. Okay, first of all, in the 80s, Carol Dweck already in her book Mindset, said there were two types. One is a fixed mindset and one is a growth mindset. And the growth mindset, that it believes that our skills and capabilities and talents and limitations are actually not set, not fixed, uh, but can be changed, is truer than ever and more important than ever. Because, uh, if we think that the skill set that brought us here is the one that will, you know, keep us relevant in the workplace, then we'll have problems. And uh, education's not helping. Corporate trainings are not really helping. And so it becomes a very individual decision. And uh, it really depends on the sense of urgency that each one has. And uh, you know, I think more evident than ever. I think it's, uh, it's happening right now.

Speaker C: I love how at the end of the day the things, you know, when what progress brings is just further mirroring the need of going back to your core principle first, principle thinking and core, um, questioning. Uh, this is fascinating, man. I think this is going to be very important, uh, you know, a very important book and topic for a lot of people and, and I'm excited to see where this goes, uh, maybe as we wrap up, you know, going back to all the changes you've kind of lived through and obviously you've had, we didn't get the chance to go too deep to your experience with, with Tinder, l'. Oreal. But you know, what is maybe an early advice you got in your career that helped you and you know, kind of worth paying forward.

Speaker A: Look, it's a funny one because, uh, you know, it's, it's more related to what I do now rather than these jobs that I did in the past and l' Oreal and Tinder, because I remember actually before them I used to work at Groupon. I don't know if you recall Groupon, of course, Shoot up to today is the fastest, uh, you know, unicorn, the company that became the unicorn, the fastest ever. The problem with that is that it actually collapsed uh, as fast as it uh, became my unicorn. And uh, that was because we didn't really focus on the customer experience. We thought that the value that was coming to the customer was based on the low price he or she would pay. And uh, my understanding throughout my career, as much as then that I applied with Tinder and uh, other companies, is that what I understood is basically that the KPIs that we look at within businesses oftentimes are the wrong ones and we focus too much on the short term ones rather than the ones that actually show the long term value of our customers and the group on experience. Uh, that actually my boss, which went against the overall management, which was sell coupons, sell coupons, sell coupons, was like, pay close attention to what happens when, when people redeem their coupons. And I was actually noticing the long lines at uh, the restaurants we were selling. I was noticing, you know, that oftentimes you would go to the restaurant that they would give you that, you know, last table close to the bathroom. And I started to realize that uh, you know, again it's not the measures that are related to your short term success but to how well you craft an experience now will pay off in the future. And oftentimes this is not really profitable in the short term because you invest money and time in crafting these good experiences. And I think we did that well. Uh, with Tinder, we were looking very much at um, you know, the usability of the app, the gamification, the retention rather than just, you know, that growth in downloads or active users. So I think that was an early tip, which is don't focus too much on the short term KPIs, but look at the customer experience which will pay off in the longer run through other metrics that we don't usually look at in our day to day. So I think that didn't work for Groupon, but it eventually worked in other uh, aspects of my work and life. So nice. Yes.

Speaker C: Awesome. Thank you for that. Well, uh, if anyone wants to pre order the book or chat with you and kind of like nerd with you on those topics where, where they can find it.

Speaker A: Sure, Sagi. I mean there's a uh, website about the book between you and AI. It's the title of the book under the AI domain then available on Amazon, available on uh, uh, all platforms out there or also on ylace, uh plus people if they want to reach out to me, just uh, independent on the book it's Andrea Yaro.com or on LinkedIn you can find me and uh, let's have a chat. Definitely you're getting your copy to your home. Of course you're not buying one. Sage, thank you so much for having me here. You're getting a personalized copy and uh, anyone reach out to me across social media and uh, my website, dandreayer.com awesome.

Speaker C: Andrea, thank you so much man. This was fun. And uh, go get the book. I think it will be an incredible asset for people in this transition. So I appreciate you coming here man.

Speaker A: Thank you so much. Sagi.

Speaker B: I hope you enjoyed this episode of Modern Business Operations. You can see the show notes and all of the resources mentioned on um, Today's episode@tonkin.com mbopod thank you for listening and be sure to subscribe for updates on future episodes. And if you're interested in staying up to date on the latest in business operations technology, head over to tonkin dot com.