Mercado Libre needed to develop a mobile app. The company’s future depended on it. Consultants and specialized firms charging millions couldn’t find the solution. Until Dani Rabinovich discovered that three of his employees were doing something behind his back. Any other boss would have fired them on the spot. But Dani isn’t just any boss. That’s why he decided to reward them. Sole Dematteo shares the B-side of the story.
Mercado Libre needed to develop a mobile app. The company’s future depended on it. Consultants and specialized firms charging millions couldn’t find the solution. Until Dani Rabinovich discovered that three of his employees were doing something behind his back. Any other boss would have fired them on the spot. But Dani isn’t just any boss. That’s why he decided to reward them. Sole Dematteo shares the B-side of the story.
S1:E2 Dani Rabinovich and Sole Dematteo: The app you get fired or rewarded for
Federico: [00:00:00] Escaladores combines two words when translating from Spanish to English – climbers and scalers. Climbers, because they climbed mountains of difficulties and prejudices and mountain ranges of impossibilities. And scalers, because they managed to scale their initial idea. An idea is always important, but much more important is to put it into practice and make it grow, even in contexts that seem adverse. Climbers & Scalers is the story of people who think they can. Reaching the summit is not only touching the top, but also enjoying competing in the most challenging league with the best in the world. This is the story of Mercado Libre told in first person by its protagonists. This is Climbers & Scalers, narrated by their own professional AI cloned voices, with full authorizations.
Federico: [00:01:14] Often climbing involves having to take some ballast off. But sometimes it's not so easy to let go. Mercado Libre had a system that gave them a spectacular result. It was fast. It was agile. But they had to abandon it because if they kept using it, they were going to die. It was that simple.
Dani: [00:01:36] Monolithic systems are very agile systems. If one wanted to start a start-up now, you would go with a monolithic system, as everything is solved in a small system. They are super agile because there is a very big cohesion in that system. There are very few layers. So, it's very agile for when you must be agile. We had a monolithic system that was very good for the first stages, it gave us tremendous agility and cohesion, tremendous control. Control and agility are key in a monolithic system. Agility in the sense that I can change things very quickly with very few people. But monolithic systems have two serious problems. The first is that they don't scale because everything becomes exponentially complicated. And the second, the more serious problem that a monolithic system has, is that not only the system does not scale, but also the team that plays in that system does not scale either. If you look at the hiring curve that Mercado Libre has, it looks like a completely flat curve for 10 or 13 or 14 years, and it starts an exponential curve since we finished this New World project. Because not only could we scale in terms of systems, but we were able to scale in terms of people. We were very, very few, until not so long ago, and monolithic systems do not scale, they are closed. That means that I cannot allow the rest of the systems in the world to interact with Mercado Libre. I am not referring to users using applications, but to other systems interconnecting, because since it does not scale, it is very risky to allow third parties to get into your system, because it is like the closest thing you can have, which is a small mistake, and you broke everything. So, we went from a very monolithic system to a loosely coupled and open system, that is, from a closed and monolithic system to a loosely coupled and open system. This is something that was very fashionable. At one point Facebook started a lot with that whole issue. So, we changed the concept of having a very closed system, played by five geniuses. We had a platform, a platform means the whole part of technology designed, open and thinking of Mercado Libre as a particular client for everyone. So, any system today can talk to Mercado Libre and us, that is, our yellow application, for example. So, this idea of creating a single platform and that we are just one more client and that the others can interact with us, saved our lives and we did it, maybe a little late, but later we would have died because we would not have been able to scale our systems, we would not have been able to scale the number of interconnections we needed. Today, our professional sellers do not operate Mercado Libre directly, they operate through systems that connect with Mercado Libre. And if we had not been able to open our platform, we would be dead, but actually we would be dead for another reason, because we would never have been able to grow the team. So, the strongest concept is, that it is very difficult to scale in time. If you think of systems that are too scalable at the beginning, you die from lack of agility, and doing it too late, you die because you don't have enough hands to do it. I think we were aiming too much at the fringe when we did that in 2010 or 2011, when we were already a very big and public company. I think we failed, but it was an apprenticeship that later marked us a lot. A concept that seems very important to me is that at the end of the day it is binary. You are either in the ring or you are not in the ring. And I think that allowed us to stay in the ring, and not be taken out of the ring. Being out of the ring doesn't mean just not existing anymore, it means not having a chance to not stop being relevant. eBay, for example, is a company that they took out of the ring, which breaks my heart because I love eBay very much and we have a very big history with them. But it was very clear, quite a long time ago, that they were taken out of the ring. They still exist and they can always make a reconversion and take Amazon out of the ring or some other competitor, but the truth is that it seems very difficult. Irrelevance is perhaps worse than non-existence, because not existing anymore, forces you to start over quickly. Irrelevance can leave you in a lethargy that is terribly difficult to get out of.
Federico: [00:05:22] Mercado Libre had to change their entire platform to be open, interconnected and allow their team to scale to another magnitude, developing all the technology they would need in the future. And all this being a public company with investors eager to see results. Time was pressing and the most logical thing to do was to cut corners by hiring consultants with much more experience. But there's a small problem with shortcuts. They don't exist.
Dani: [00:05:48] Soko left Meli in 2007 or 2008, around there, and there were a couple of years where there was no CTO, and I was there acting but not appointed. I was there for two years, half that was acting, but no, I wasn't appointed. And it was obvious to me, that our platform that got us there wasn't going to get us to the next place, and at some point, you have to change this. But it was a very radical change, so it was a little bit vertigo-inducing. And what Ground Zero or New World capitalized, I mean all the code is written as GZ, Ground Zero was the code.
Dani: [00:06:14] What capitalized on that was a conference that we went to with Marcos, where we heard people from Facebook speak, who mentioned that that I was talking about earlier. They see Facebook as an open platform at the time, like the content at the time was completely open. Facebook's API was revolutionary because they really thought about authentication systems, and they thought about millions of customers. It wasn't thought of that way before. Before the system interactions were like all very business B2B enterprise. As is almost always the case in life, systems that don't have to compete are bad. So, if I make a deal between a company and another company and then the system is found, but the deal is already done, all the chances are that this system is awful. It is like the system to collect taxes, as you must pay taxes if you do not go to jail, the system is very bad because it does not compete. The systems that are good are the ones that compete. So, Facebook set out to make an open API and basically had to achieve millions of integrations, which was very good.
Dani: [00:07:07] And at that moment with Marcos we said look, it is time, we have to rewrite everything. You can't make an API based on what we had. Of course, there was always the little company that gave you the hack, the shortcut that opened the API. And as we always say, and it's a phrase that I use very much to close the talks that I give sometimes, and maintained for many years, was always the same phrase which is, that there are no shortcuts to go anywhere worth going. In fact, they recently sent me one of those stickers on WhatsApp. They put my little picture below and it says no shortcuts. So, when you see that in a discussion someone wants to put duct tape on something, the other person replies, you know what, there are no shortcuts. Okay, duct tape, it's okay to score a goal with your head in the ninety second minute if you must, but for these things, there are no shortcuts. So, we said, we're going to write everything back and we're going to write what's already open. And the catalyst was that conference, which unfortunately I don't remember what it was. Part of the logic of creating a platform, is that an API, is this idea of having a non-graphical interface. But how to be able to interact with systems and that the client, the customer in technical terms, is what device you're using to access. A client, for example, could be a PC, a client could be a cell phone, or a client could be a laptop. The customer had to be agnostic to that. So, this foundation that we laid in technological terms had to have N clients. For example, one client had to be web browsers, a different client had to be mobile applications, other clients had to be third-party applications. So that's exactly what the mobile world was, a parameter of that system. We had our strong mobile application quite a few years after that, but we already, of course, had the idea of doing it. It was our own decision, what we talked about timing. In 2007, 2008, it didn't make sense for us to do a native application. Every three minutes someone would say, I read all the newspapers. But this is not the time to do it. The systems were very bad, the programming language was very bad. We only started to make some systems a little bit later and I think we were, between quotation marks, late adopters, by our own decision.
Dani: [00:09:09] In all honesty, I'm going to tell you something that I think we did very well and something that we did very poorly. I'm going to tell you first what we did very well. The first thing is not maximizing time to market. As we are still saying a lot, the thing to maximize is time to impact, not time to market. It's completely unnecessary and irrelevant to get out fast if you can't scale fast afterwards. It is very common. I don't want time to market. Great, but then you are irrelevant for five years, and the truth is that it is not funny at all. So, time to market for me as a concept, is very, very, very little relevant. The question is how to achieve scale as quickly as possible, and sometimes one way to do that is to wait. So, we didn't do that by choice. And that was very good. It made us learn; it made us not necessarily learn things that were too impractical to use at the time. That, I think, worked out very well for us. The one that we did wrong, is a funny anecdote too. It's that at the time, we conceded, and we did what I say all the time not to do. Which is instead of learning something, not to take any shortcut, and face the issue. We hired some consultants to start helping us and help us make the application, because in theory there were some Brazilians who knew a lot about mobile applications. Well, we went to Brazil, we talked to them, and I said, they really know. That's it. We asked the consultants to make the first Mercado Libre application. Big mistake. Basically, because we are not learning anything. It was obvious that even in a moment, even when they released it out, we were going to have to throw it away in a moment and be everything slower. But that was a mistake I made quite a bit.
Dani: [00:10:39] The good thing is that this error had very little impact on life, but not thanks to me, but thanks to three people from Mercado Libre. Because in a moment a mobile application comes out, an X mobile application comes out. Obviously, we had an open platform. That means that anyone could develop on top of us. They launched a mobile application, which was not called Mercado Libre, I don't remember now what it was called, but you could operate on it. And it was ten times better than the one we had. And in a meeting, I remember I was strongly complaining, saying, how could it be that someone could make a better application than us. And then, one of the guys came up to me and said, the truth is that it was us. It was made by three employees of Mercado Libre, Mati, Sole and Martín. The three of them.
Federico: [00:11:21] You must be connected with the best in the world to learn and be encouraged and do things at the right time to maximize the impact. And it's also important to be connected to what's going on at home. Dani discovers that the mobile app they were struggling to develop already existed and worked very well. And not only that, he discovers that it was made by his own employees, outside of Meli. And what do you do with employees like that?
Sole: [00:11:36] Well, it was amazing that moment because we were at Meli building our platform so that other developers or even companies could develop technology mounted on top of ours. And at that moment, with a partner, with Mati di Gregorio, we were eager to learn other things besides that. And to live the experience of doing something that we came up with ourselves. To have a little more freedom. I remember the first thing we did to test how we were working together, was a few years before, a kind of site where you could place bets for the World Cup. I don't remember what year it was. Then, we made an internal site, so that at least the work groups could make bets in a closed group on how the matches were going to turn out. And that was super fun. And after that we decided that we wanted to learn mobile, specifically for iPhone, which was what most caught our attention. One day we said, well, why not, when we leave work, we go to my house. And that's how we started. Until one day we were thinking, thinking and thinking, and we said Hey, it doesn't make sense to do what we are doing. Meli doesn't have a native iPhone application, we are building the API, why don't we do it ourselves? And while we learn mobile, we look at how we are building the interfaces inside, because maybe what we are doing from outside doesn't work as well or there is something to revise. So, it was useful for us from both places. And that's how it was. We started. At the beginning we were a bit afraid to say that we were doing this after hours, because it was not very well regarded if you had a job outside of Meli or a business venture. It was like, hey, he is going to leave. So, we decided not to say anything. So, we didn't say anything. And we started, and some time went by, and we started to work on it, and we got into it, and at one point, we had an application that worked very well, and we put a lot of head into making something that, although we were not designers, the design was not very aesthetic, but it was very thought out, so that the application would flow, that it would work fast, something that at that time was still a little difficult to do. And it worked very well. We also asked Pancho for help with whom we had worked together a few years ago. And we added one more partner, to make the Android version, as we didn't know Android at that time. We said, hey, we need someone else, and we added him. And we started working every weekend two or three nights a week, programming until 3 in the morning. Mati would stay, as he lived far away, he would sleep on the couch in my house, and we were very excited. We loved doing it, and it also helped us a lot to see what we were building inside Meli. And then we would go back to the office and say, hey, what we did here is wrong, it's not understood, it doesn't work, there is no applicable use case. It was like it was useful for both sides. And one day we uploaded it to the store and a lot of users started to download it. And the reviews, you saw that you can give a rating to each application, the reviews were very high, it was great. I think it was a double effect, that the application worked fast, that there was no other application to navigate in Mercado Libre, there was no alternative. And one day in a meeting of a quarterly update that Marcos was holding on the Market Place, at one point, they were talking about mobile, and Marcos made a comment: has someone seen this application? And I was there, I was hard, and I said that's it, I'm not getting out of here alive. I hadn't said anything at that moment. Then he said, hey, it's hot, why don't you investigate it? He just threw out the comment. I left that meeting and I talked to the guys and said, hey, we must clean up our act. And that's how we talked to Dani Rabinovich and from then on it was all spectacular, because in the end, Meli ended up buying the application from us. We were employees. I mean, it's all very strange. And it was like a mini revolution inside too, because, how crazy, what a crazy story that, being employees, they bought something that we had developed, that even though it was in our free time. And after that, which was great, they gave us the responsibility of building the mobile team inside, and then we started to have some life, and to be able to live on weekends without having to be programming because we started to build a team inside Meli and that's how it came about.
Sole: [00:15:26] The truth is that their first reaction was, it's great what they did. Too bad they had to do it on the weekends. Let me see what we can do. There wasn't much more than that. Luckily, maybe the comments Dani had in mind, he told them to someone else. Luckily, he didn't tell us. I think it was great because I think that, beyond having done it, we had really put a lot of thought into it and we had made a lot of decisions that even at that moment, going to the technical side, there was a super polarized discussion between building native applications in the native code, the operating system that was for Apple, or making what was called hybrids, which is another technology. The hybrids were theoretically less expensive from the development side, but less performant and the native ones performed better and, in theory, it cost more development because basically you must duplicate your code. You make the code for the application and the code for your website. Well that was the polarization of the discussion at that time. We decided to make it native, at a time when Facebook had decided to go the other way. At the time, when they decided to buy it, there was a whole debate about whether we had to go that way or not. We were very firm that it had to be native, we fought with a lot of people, a lot of senior people. And the argument was that Facebook, which is one of the biggest, is going the opposite way. And we said they're going the wrong way. Within two years, Mark Zuckerberg came out and said that they had indeed made a mistake in the decision to have gone to a hybrid app and they were doing it back, yes, native.
Federico: [00:17:02] Let me tell you, you can learn a lot more from a mistake than from an expert. And of course, you can learn a lot from your employees, even if they have a venture outside of the company.
Dani: [00:17:10] We bought them the application and they became the heads of mobile, which didn't last long, because later, it got out of proportion and it didn't make sense for a group to do mobile things, it is like a group that does web things. It was a platform. Everyone had to do mobile. We made a big effort to build the mobile team and then a big effort to destroy the mobile team and there is no mobile team because it's like there is no innovation team, it doesn't make sense. And each of these guys then went on to specialize and do other things. The real value is to build things, not to talk mobile. So, this gringo thing, I'm not an expert in this, why am I going to do this other thing. The value of this group of people is that they can tackle problems, not that they were experts in this. In fact, it's much easier to hire an expert than someone who can tackle a problem and solve it. So, I think, in retrospect, I'm very frustrated by the decision I made at the time but, on the other hand, I quickly let myself be convinced. It's horrible, it's very present in my mind. That's why afterwards, at other times, I think I didn't make that mistake, and I think we had, as we often say, the enormous luck that these guys were dedicated to doing that, instead of watching Netflix. Sometimes it's hard to understand the forms that luck takes.
Federico: [00:18:16] Obsession helps to become good at everything you do, from managing the tensions of the threads of a magic table to the attentions of the people in your audience at Meli.
Dani: [00:18:27] I think it was an interesting journey. In fact, being the head of technology was not the last stop. In fact, today I have a different responsibility than being the head of technology. Greater, I think. I'm responsible for the product, and I'm also responsible for the customer service and fraud prevention group. I think that, in essence, I am responsible for the areas that are more cross-functional, closer to the product. I would almost say, I'm responsible for product today if I had to sum it up in a broad sense. Customer services is a very central part of the product, although there is of course a big part of operations. With respect to the progress of working very much in the technology part, doing other things. I think that what happened, naturally, is that many of the things that interested me, were things that I did with a lot of passion at another time in my life, such as teaching or, for example, doing magic. They started to play in my favour when it came to changing roles because of some basic issues. Going to a table twenty-five times every night, gives you a very big gymnastics of reading people. There is a very, very famous magician who is still alive, maybe the most important magician, one of the most important magicians in history, and definitely the most important magician alive, his name is Juan Tamariz. He is a Spanish magician. He's kind of like an elf on earth. He wrote quite a bit. And one of the things he talked about was this thread thing. I'm pretty sure it was him; I hope I'm not confused. This idea that if you have an audience, you must have a tight thread with all of them and that a lot of times the threads are kind of loosening up. So, if I look at you, the thread just loosened with you, because Fede stopped paying attention and looked at his phone. I realized that naturally you did that, but at this moment I don't have my objective on you. I would have done something to make you pay attention at this precise moment, because it is something that must happen while you are talking and generating. In other words, there are several layers in a conversation. There is what you are saying, there are the meta messages, there are the different groups that are listening to each message and each one decodes it in their own way and all those things that happen. Apart from all that, there is the tension of each one of the people who are listening. And I trained that a lot, for example, teaching. You realize the kid who is leaving, the one who is coming. It is this kind of things, which are gymnastics acquired over a long time, when you must become a speaker, for example, which was the stage where we went from being twenty, thirty to one hundred, two hundred, three hundred, you do not have time to put those gymnastics because it is a big mess, and you have to take care of a thousand other things. So, for me, having done a lot of things before, such as being in contact with audiences, I don't think I would have been able to do that if I had to go to the university to develop that or anywhere else, coach myself or whatever. The same thing, a very important part of a leader's job is to be a great communicator. I'm not saying I am, but I did have a lot of training in that. Then dabbling in softer skills helps a lot. I would hate to draw a parallel because obviously if I get to say, connect the dots, somebody is going to say I'm comparing myself to Steve Jobs and that's a crime. I'm not saying that. Please nobody ever make the slightest parallel to that. But it definitely exists. I'm not comparing myself to Steve Jobs at all. What I'm saying is that it definitely exists. When you do a lot of different things, it's rare that they don't work. And it helped me a lot.
Dani: [00:21:28] And as a general reasoning, I think that the transition to generalist is a problem that is not very well solved in Western culture, in the sense that you ask someone to be a super specialist in order to be able to enter a company and shine in some aspect. Then there is a transition to generalist that is not solved, because if I did a PhD in something incredibly technical and I spent ten years studying that, and then I spent several years of my life exploiting that. By the time you tell me to lead a team of people, I don't have a tool in the box. So that is not very elegantly solved. But being Argentines, we are a little bit lucky, we are not so many specialists and of course in Argentina we are all a little bit of, sort everything out, but it is a very natural problem. And in fact, we still have these things of, do you like the technical career, and you like the manager career. We are working hard to close this concept so that you don't have to become a manager to progress as well. There is also the reverse, which is people who love technology and don't want to progress by leading people. They feel they are at a point in their career where they say, oops. It is like, if you are still programming by the time, you are forty years old, you are dead and that is very serious. So, we are doing a lot of things so that good technicians can do what happens in other parts of the world, where they are stars. There are good and bad programmers, there are super senior and super junior programmers, and super junior and super senior managers, but a manager is not necessarily better than a super technician and it is something that is a bit in the vox populi that anyone who leads is better. And no, that's not necessarily the case. So, I think to close the concept, almost anything that you do, or you study in depth very well, it's very likely that it's going to serve you at some point down the road.
Federico: [00:23:18] In the next episode of climbers & scalers, Dani is going to explain what the tango El Choclo has to do with the way he works and motivates his team.