Prosus N.V. (PRX.AS) Earnings Call Transcript & Summary

Hello, good afternoon to everyone, and good to see you, and thanks for joining us. Today, I'm here with my friend and the CEO of iFood, Diego. Diego how are you doing?

I am great. [indiscernible] yes, which means I'm always improving.

I'm super excited for this call. As I told you earlier, I put on my special jacket with my fancy shirt with blue buttons. So that -- so you know it must be an important one. But here, today, we want to talk about iFood and the growth at iFood. iFood is growing at 30%. It's profitable. The restaurant business has 20-plus percent margins, over 100 million orders. I'd love to get a sense from you today, Diego, of kind of what's driving that. And maybe we can double-click on your subscription product, Clube. So why don't you take it from here?

For sure. Thanks very much for the invitation, Eoin. I'm very glad you are well dressed to have me here. But unfortunately, I'm the best I can do. So I'm sorry if I cannot be in the same level that you are, just wearing a small -- a simple shirt. For those of you that don't know, we are in the middle of the Carnival in Brazil. And this is like the largest popular festival that happens. Just for you to have a sense, in Salvador, in the Northeast of Brazil, there are 11 million people that go to the street during 4 days to have fun and to celebrate the Brazilian culture. And iFood, by the way, it's the main sponsor. It's the main brand that leverages the Carnival to show how we are a company that understands the Brazilian cultures and very fits the Brazilian routine. So big pleasure to know all of you. Let me show you here a little bit of Clube and how we are building an ecosystem here. So first of all, it's important to understand the history, the story of iFood throughout this year. So we are a 12, 13 years old company, very young. We back to 2011, when the company was created inside this start-up called Disk Cook. And Disk Cook at that point in time was basically a call center, a company that was mainly trying to make the ordering of pizza easier. So in the past, you just get your phone, it was busy most of the time. You couldn't talk to the restaurant and so on and so forth. So basically, they got all the menus, they printed the magazine, they send it to the neighborhoods. And then people started to call to the call center that sent a fax to the restaurant and then the restaurant delivered the pizza. So that's how we started. That's how -- that's where iFood was created. 2 years later, Prosus joins -- sorry, Prosus joins the cap table of iFood through Movile. And a few years after we reached the first 1 million order per month, 2015. It's only 10 years when we reached the first 1 million. 3 years after, we launched our 1P model. So basically, everything that you know from the companies that operate in the country where you sit, it's basically what we evolved here, but we did it in a different manner. At the same time, we launched our initial -- the first proprietary AI models that allow us to not only grow very fast, but at the same time, prove the profitability of this model quite soon, sooner than most of the competitors worldwide. 2021, the company launches the adjacencies. So groceries and fintech started here as very small investments that we did at that time. Many people at that point in time were asking us why we were -- it took so long for us to launch this business? And the answer was simple. We love focus. So everything that we did until here was focusing on food delivery, basically. 2022, we reached the breakeven. We launched the ecosystem vision 2024. A few months ago, we reached 100 million orders per month. So after 10 years, we go from 1 million order per month to 100 million orders per month. And we are just getting started. So we don't see that as a great number. The great number will be 10x this one here. And now we are just building the path to go now to the 200 million and then to the 300 million and so on and so forth. So today, I want to break down the iFood to a more specific business that is our loyalty program. So initially, the question is, why do we have a loyalty program? So first of all, when you think on how we can grow in terms of order, this is a basic math that comes from the number of users that really comes an order, multiplied by the frequency. So we knew that growing as fast as we were back in 2017 to 2020, we would not be able to continue to speed up the orders growth only with customer acquisition. And the reason is simple. We are more and more penetrated in the Brazilian houses. And at a certain point, the growth of new users would decelerate. And that's what happened. So as you can see here, we come from 83% growth in this period to 20% growth in this period here. But at the same time, we said we need to start to build a product that would allow us to compensate this huge penetration here on the unique buyers and balance it with frequency. So when you take a look here on the 2017, we had around 2.7 monthly frequency on the food delivery business. And now we are already at 4.1. So the whole idea here was to build something that could allow this to happen and as a consequence to reach this. I told you that we reached the 100 million orders in 2014, and that happened on August. Actually, we end 2014 -- I'm sorry, 2024, not 2014, with 150 million orders. So the combination of huge penetration and building a product that would allow us to increase frequency in all the cohorts of the company gave us this result. And this is what I want to talk about. I mean, how we were able to do something like this. So I want to talk to you about Clube. Clube in Portuguese means something like club, like reunion, like being together, being in a place where you can find many things. So that's the rationale behind the word Clube. So Clube is our loyalty program here at iFood. We had this idea of starting to work with that back in 2019. In 2020, we started what we call the growth mindset phase. What does it mean? At iFood, everything comes exactly to these three steps, everything. So we run several tests that we call internally Jet Ski, which is basically a testing machine. So we have a way to do. We organized a product and engineering at iFood that allow us to reach results like this, like only at Clube and this year here, we were performing more than 1,000 tests in a single year. So as you can see here, we are not thinking about growth at this point in time, we are thinking about growth mindset. Why? Because I don't know the solution. I know the problem. I need to test like crazy, as you can see here, to find the solution for all the time. In the beginning of 2022, we believe we reached some solutions that could give us the right product to the problem that we were trying to solve. And with that, we went to the second step, which is the product adjustment in order to monetize, in order to balance, offer demand, in order to balance all parts of the ecosystem. And then we started to increase our growth here because more and more, we were able to reach the monetization results that we were looking for. We call this a boat. So we come from the Jet Ski, the boat concept. You are bigger, you are well suited with other tools that allow you to do much more things than a Jet Ski. So that's what the boat represents. And as soon as we find a way to monetize this as we need, not only for us, but also for the merchants that is part of this equation, we go to the growth hacking phase. Now we know the playbook, it's time to accelerate. So as you can see here, we come from 9% of food delivery orders coming from Clube to currently 39% of the orders coming from Clube. So what does it mean? It means that in the beginning, I was looking in this growth mindset phase for a consumer value proposition. I was trying to validate that. I have flexibility to test several models. I did more than 100 segmentations in my customer base in order to find the best fit for the consumer. When I go to the monetization, what I'm looking for is the merchant value proposition. Why? Because the merchant is very important here in order to incentivize the platform and as a consequence, the consumer to acquire the Clube program and as a consequence, using it. So we built a platform for merchants to invest, and of course, the value proposition of the merchant, it's totally linked to the idea that they have a great LTV, a great ROI. There's no other way to make this work. I use it to say that everybody that says that has a loyalty program, the first thing I say is, let me -- show me how many clients actually use, show me how many orders are coming from this. If these numbers are big, it means that the consumer loves and the merchant loves as well. Without this, it's impossible to have a huge penetration of a program like this. And the third phase, which is the growth hacking, it's when we started to find mechanisms to make the program even more important for everybody in the ecosystem. So for example, recurring membership instead of just a onetime monthly payment, unlocking new verticals, it's another way, creating a retention engine through a subscription model makes our customers the best ones and the ones with more retention go to another level and as a consequence, represents growth for us. The way we do that, 70% higher LTM user retention, customer acquisition, given higher LTV, which means because we have results like this, we can just speed up the process of acquiring customers. And again, as a consequence, having 40% of the orders of iFood coming from Clube. So in the end, where is supporting the Clube here? So first of all, it's an extensive use of technology and problem solving. So this is not about like having somebody that is like clever than the others or somebody that found a code that nobody else in the world thought. It's not about this. It's about technology and problem solving. And it's about growth mindset using these two tools. How we did that? So first of all, intense user profiling with hyper-segmentation. So we know exactly each customer and each restaurant, and we find the right match for both of them in order to make the consumer happy and at the same time, a good LTV, a good ROI for the restaurant. Second, high scale testing. As I already told, we need to have like a problem-solve engine to make you find the best solution that is possible, and not the solution that you envisioned. Third, mechanisms. We have like several mechanisms. And of course, I'm writing here coupons and discounts, but actually, the mechanism is much more about how the product works, how the UX works, how you understand what people are looking for. And as a consequence, you segment everything to have the best results for the consumer. So as a consequence of these mechanics, we have currently 80% of member orders with at least one Clube benefit, which means people are using people like, people understand. And finally, we have what we call a concierge AI. So we have like huge and deep and profound models that knows exactly this user mechanics fit recommendation that I mentioned before. And this incentivizes optimization to increase conversion on the side of the consumer. And at the same time, it makes the investment of the restaurant much more rational, which gives consumers and restaurants a good balance and make them participate in the ecosystem. What we are envisioned here. So first of all, maintain this growth hacking mindset. We are a company that bet believe on culture. So betting on maintaining growth hacking mindset is part of our culture. We cannot lose that. And this is what make us drive the core business growth. So even though we are growing in the other adjacencies, we are betting we can continue to grow the core business at the highest level that you can see in the world. Higher membership penetration on the food delivery. So we still keeping this, and we believe we can go higher, leverage members, high engagement to support adjacent vertical growth. So as soon as we have been very successful on Clube in the food delivery, now it's time to roll out this to the new verticals and support the growth on them. We already have Clube penetration on grocery to 46% and pharma to 31% orders. We started this test 3, 4, 5 months ago. We are already with this penetration here in groceries and pharma. We are now adding other verticals here such as beverage and convenience, among others, such as pet that will come. And at the same time, we are increasing the penetration in these verticals. And finally, leverage members, high engagement to enhance membership value proposition. So the stronger this program is, the higher the possibility to increase retention and make an ecosystem a differentiator comparing to competitors. Finally, huge opportunity to cross-sell partnerships, not only in the retail media space, but also in the Clube space when I bring more companies to participate in this program. So now what we are focusing is bring all these categories to the Clube. As I told you, groceries, pharmacy is already there. Now we are bringing beverages. And next step, dine-in store, iFood benefits, which is our meal voucher business and our wallet. The next step is that Clube Everywhere. So we not only bring everything that you have, such as the food delivery and the other adjacencies, but also OLX in Brazil through the classified business. Despegar, if everything is confirmed and Despegar becomes part of our Prosus group. Sympla, which is -- if you are in U.S., is the event bright of U.S. So Sympla is a ticketing online business that will also be part of Clube. We can bring mobility. It doesn't mean we're going to invest in mobility. It means that this service can be part of it, and we are in intense negotiations to have one of the main mobility business in Brazil as part of the Clube. And finally, the more and more we grow this, the more and more we bring ads, retail media, the more and more we're going to have our fintech business through payment and credit as part of this program. Thanks very much. I hope you were able to understand what drove iFood's growth, balancing user acquisition with frequency. I hope you understood that as part of the results, actually, we have a backbone that is made of culture, that is made of growth mindset, that is made of problem-solving mindset, that is made of testing and finding solutions through test. And as a consequence, delivering these great results that you saw. Thanks very much.

Thanks, Diego. That was amazing. So Luke, can you help us with directions on how to ask a question, please? And then I'll ask a couple of myself.

[Operator Instructions] I will now hand back to your host, Eoin Ryan, to take your questions.

That's great. [Operator Instructions]. Diego, for me, Slide 6, is one of the most interesting slides because what you have for me in iFood is an absolutely amazing consumer product. But what Slide 6 says to me is it sits on a bet of automation. And I love what you said, high-scale testing. The Internet used to be about, oh, we A/B test. But it seems like you did the whole alphabet at once. Do you want to -- can you talk a little bit about how you test and how automated that is and how that takes you from a ski to a boat?

For sure, Eoin. So the way we do that here is, first, we built the entire architecture of the company to be -- we say levelized it, means it's like a level where everything can be detached and added to another piece. You could say if this is an API-based company, that's a way to say this, but it's more than this. It's also preparing the company, which means the people to have access to everything and to use it. So let me give you one example. Our data lake here is 100% clusterized and accessible to everybody in the company. So at iFood people, 75% of people, they called using SQL. I'm saying 75% of people, which means people in marketing, finance, operations, marketing, PR, everybody. Everybody knows how to code on SQL and they go straight to the data lake to do correlations, to understand hypothesis and things like this. So the entire architecture, as I said, is prepared to allow people to have this flexibility. That's the first thing. The second thing is we don't treat people here as corporate people. We treat them as entrepreneurs. And entrepreneurs, they have to have the gut to go there and do something. They have to have the gut to go there and test what they believe should be tested. So what I'm trying to say is that -- and it's so far from answers that I listened from other people. What I'm trying to say is that our selection program, the attraction program in the HR department is also linked here because if I'm able to identify people with this entrepreneurial behavior, actually, these people combined with this aperture, it's what allows me to deliver this kind of results. So in the end, because of these two components, I can do several tests every single day because people have the tool and have the behavior. The last thing, Eoin, is we are very, very product-oriented. So what I'm trying to say is that iteration is something that is part of our behavior. So we don't build nice PowerPoints. We like to get PowerPoints without any like colors or things like -- I don't care about this. I just care about what's our vision, what's the problem, show me the tests. That's this. So we don't have like innovation committees. We don't have like this kind of stuff where people get -- stuck because they need to wait for the paid to show the results and so on and so forth. People have the freedom to -- as soon as they understand the results and that we can go through the steps that I showed, they just can go. They are free to do that. And then people will ask themselves, but I mean, this will create problems in our company. If you don't have culture, yes. If you have a culture that is followed by most of people, problem will be easy enough to be fixed.

Okay. I think I probably have a few more follow-ups on that, but let's go to the phone lines. And Cesar, you're first up. So remember to unmute yourself. Cesar from BofA.

Very interesting, and I'm from Bank of America. I wanted to ask about how this ecosystem in your view, impacts the profitability of the -- of your business. If you can talk a little bit about it.

Yes, for sure. It impacts positively. And why? Because the whole idea of the ecosystem is to have the right balance of incentives to organize offer and demand. So if you remember, in one of the slides, I mentioned that we built an investment platform. What does it mean an investment platform? It means a place where the merchants go and they insert a certain level of investments and they indicate the level of return that they are looking for. So first of all, there is an alignment of interest here. I'm not just saying, give me your money, let me see what I can do for you. What I'm saying is give me your targeting in terms of returns. When he tells me that, I can tell how much I can penetrate or not. The more you invest, the more I can give you something, the more you want a high returns, you will find less customers and so on and so forth. So there is a clear vision on how to perform investments for the merchants. On the other side, I mentioned to you as well that there are several proprietary AI models that understands one by one customer in one by one restaurant. Why is this so important? Because the way you set up your UX is to show the best customer to the right restaurant and the best restaurant to the right customer. If you are able to do that, you can deliver what I mentioned here in the beginning, the investment platform. And if you are able to match that, what happened is that in the end, the restaurant will grow with a nice return, the customer will come back. Where this -- where everything ends, ends with the customer with a different behavior, liking more and more a specific restaurant and organically coming back to it. And this close the entire profitability to the restaurant and close to us. So I can tell you one thing. We are approaching now the profitability of per order of Clube to the organic order. So what I'm telling you is and see how crazy is this, the profitability that I have for an organic order is getting closer to -- I'm sorry, Clube is getting closer to it, which means that at a certain point, and I hope this year, we're going to have a preference to acquire through Clube and not necessarily organically, meaning I want to have more and more of my organic orders and customers inside Clube because there is no trade-off in terms of profitability. So that's why the company -- and that's the way to make the sanity check is how iFood was able to grow as fast it grew and at the same time, increase profitability. It's because of what I just mentioned, once Clube is 40% of the orders.

Can I ask -- yes, one quick one, one last one, if that's okay. I wanted to also ask, I mean, not looking for a very specific guidance, but conceptually, how much do you think you can drive growth in order of frequency in a couple of years?

Well, conceptually, what I can tell you is my adjacencies is still very small compared to the food delivery business and compared to the level and compared to the online market that already exists. What does it mean? It means that there is a huge chance of having a good incremental frequency coming from these adjacencies. Remember that pharmacy, groceries just started Clube like 3, 4 months ago. And I can tell you one thing, it's already much better than we expected in terms of frequency increase.

That's an absolutely stunning stat that Clube is coming up to organic in terms of profitability. Just one point of clarification. When you say matching the best customer to the best restaurant, is that just done based on extreme personalization and what converts the best?

Yes, exactly, Eoin. So the way we organize it, our data and our decision tree and our app is -- was in a way that I know exactly who you are. So what I'm trying to tell you is, if a restaurant made a certain investment, and this restaurant sells or cooks, I don't know, Brazilian food and you were looking for a pizza, which is not a Brazilian food, even though we eat a lot here, it doesn't make sense for me just to push and push and push and make you get this coupon, get this voucher, get this promotion and so on and so forth. In the end, you may say, yes, I'm going to eat. But when I look to your profile, it doesn't mean that you are looking for it or you like that or you're going to change your behavior towards this cuisine. So what the model does is, instead of just trying to push everything, it says, I know what Eoin is looking for more or less. So let me tell you one example, Eoin. I could have here incentives from pharmacies straight to you. Or I could just say, I know that Eoin now is searching for something that is related to the fact that maybe he's just a father right now, from one day to another, he starts to search things that gives me this idea. So what do I do? The model instantly tries one or two things to confirm if now you are a dad. And if you are a dad, then the model starts to work with the incentives coming from pharmacy towards you. But if you just went there because maybe you were like just buying for your nephew or cousin, doesn't matter, the model recognized that and stop to send to you because -- I don't want to just push. I just want to just make you make an order. I need to balance this. So this like their individual personalization, not only on the consumer side, but also on the merchant side, it's the backbone of our architecture here.

That's super helpful. I was getting a little bit worried there when you were going to give examples of what I might be getting from a pharmacy, but thankfully, you didn't go there. Let's go to Silvia from Deutsche.

Interesting presentation. And I just wanted to ask about how more these findings can be leveraged within the Prosus Group, in particular, Diego, just wondering if all the testing that you have done could be applied in other countries potentially? Or is it very much tailored to the Brazilian customers? And is there anything in particular that you can tell us to think about what are consumer habits when it comes to ordering food on demand in Brazil?

Great, Silvia. So I can tell you, and I'd like to be very precise in many examples, Silvia, to not just give like conceptual generalist answers. So actually, when you ask, can we use that in other countries, there are like several assumptions before that I need to tell you. So first of all, we are more and more working as a group. So very simple example. Everybody, is on Slack. And you may say, okay, but that's so simple. No, it's not simple. I mean, pushing all the companies to the same communication platform and making people start to communicate their form groups, exchange and et cetera, it's just a very simple example on how you start to change collaboration. So that's the first thing. The second thing is Prosus has a central team that understands and work with the companies towards AI. So everything that I mentioned to you here, even though it's not 100% AI, it's on top of models. So the product and the mechanics and et cetera, are on top of AI models. So Prosus is aware of it. And because they are aware, they can just grab two different hands and say, okay, you have to talk. So I want to give you an example. How many times I went to China? Several. How many times my team went to Romania where eMAG sits? Several. How many times OLX from Germany and Poland came to Brazil? Several. So this last like 12 months have been very intense in terms of how we connect and how we exchange. That's the second thing. The third thing is from a technical point of view, this is totally easy to be replicated in other countries. What you need, it's time to gather data from the specific market that you're looking for to train the models. And this depends a lot. So in places where you find a great correlation with Brazilians, this training will be much faster. In places where people are totally different, the training will be slower, but it will end up in the same place. So part of the thesis that Fabricio and Ervin are communicating regarding jet, for example, is based on this, on the fact that we can replicate these models, replicate this engine and as a consequence, find good alphas in the new way we're going to do in the companies that Prosus will put in the portfolio.

Thanks, Silvia. Great question. Fantastic answer. I mean it's effectively -- regardless of culture, these models identify human activity and tailor the product to it. Next question is from Andrew at Barclays.

I guess my first question is to kind of come back on the history of Brazil and some of the competition that you guys have seen from Uber Eats and Rappi in particular. I guess you've touched on some of the factors that helped you kind of fight them off. But I just want to be clear on what you think the key things were that kind of enabled iFood to win? And then if you kind of extrapolate that thinking to other countries potentially with Jet, kind of what you think others have got wrong but you guys can add value to around kind of defending those leadership positions. And I guess as a follow-up as well, I would be interested in what didn't go so well in, I guess, places like Colombia, Argentina. I know those were small. I know it was a while ago, but kind of curious on just the learnings as part of that.

Great. So Eoin, I will start here. If I don't cover all the points, just let me know. So first of all, we have intense competition here in Brazil. Even though the company is a clear leader, it doesn't mean that we didn't have this necessity to fight and fight hard. So through all the time, we had Rappi here. Actually, Rappi is still here. It's a Colombian start-up. Uber Eats left Brazil after several years. Glovo from Spain left Brazil after several years. We had the -- one of the main retailers of Brazil, Magalu investing in a competitor many years ago, like 2020 or something like this. We had [indiscernible], which is one Brazilian acquirer that did the same and so on and so forth. Why were we winners until now? Because we are entrepreneurs, because we don't rest, because we love technology, and we love to follow up on evolutions, and we love to lead what happened in the country. We started working with AI here in 2018, nobody was talking about this here. We have currently more than 150 proprietary models. Our models that run the entire company, everything done here, everything done by these people here that back in 2018 didn't have anything. We don't follow. We are leaders. We go everywhere. We learn from people. I go to China every year since 2018. I go there sometimes twice, three times a year. I go there. I spend time. I get a motorcycle, and I do delivery with everybody there. I go to a restaurant and I operate a restaurant in China. I do that in U.K., I do that in U.S. We are entrepreneurs. We act and we react. When the entire market were giving like money for free back in 2018, '19, '20, '21, we had companies coming here with $500 million -- $1 billion in their pockets. We don't fight with money. Money is not a problem for us. We are a very profitable company. We have a very sizable shareholder. That's not the point, but we don't think we've money in our minds. We don't think with product. We think with product mindset. We go to the street, we live, I visit and I spend an entire day per month with restaurants, an entire day per month with drivers, one entire day per month with consumers, one entire day with supermarkets and pharmacies, four days. If you have 22 business days in a month, it's almost -- it's almost 15% of the business days. So that's why we win. That's why we win. We don't have problems to face innovation. When we went from the marketplace to the 1P business, most of the companies in the world that is starting the marketplace, the 3P business, they suffered to do that. Most of them lost their leadership. We didn't. When we had to understand how to make the other adjacencies, we did the same, and we're already performing quite well. Our fintech in Brazil, it's already performing better than the banks that serve the entire segments of the economy. So that's how we win. So what am I afraid? I'm afraid of losing my culture. If I don't lose my culture, I will win. If I lose my culture, and it doesn't mean we're not going to suffer. It doesn't mean that we are always the strongest. It doesn't mean that. It means that we are always on board. We are always fighting as entrepreneurs. Eoin, I thought -- I didn't get the other question.

One of the other questions is perhaps some learnings from Colombia.

Oh yes. So we went to Colombia back, I don't remember like '17, '18, and we were like going quite well there. I mean we were like a 40% market share. We went from nothing to 40% market share. But at a certain point, we faced a very good debate here. The market there was not growing as fast as we want, the market in total. So we just said, okay, if I launch a new adjacency here, I mean it's going to be bigger than Colombia, like in 12 months, in 24 months. And we said, I mean, what's the point to split my focus, to split my team, to split my platform, to have the language. If the market there, the entire market of the online food delivery will be smaller than a new adjacency. So that's what we did. So if you remember, back in 2021, we launched the new adjacency. That's when we said we're not in Colombia anymore. So the reason is this, it's in the end, Brazil, it's so big in Latin America that we prefer to focus. And if you remember, in the beginning of the presentation, I said we love to focus. We love to focus. We can go there, we can test, we can perform. But in the end, I love focus. So we prefer to do this, do the test. What are the learnings? I would say the main one is something that most of you already know, it's a local business. This is not a business that you can get something and you apply like an automaker, right? The same car you sell in Germany, you're going to sell in Brazil, you're going to sell in United States. The brand can be the same everywhere and so on and so forth. What it's not local, it's the way you organize data throughout the geography. So if I was in, I don't know, in Europe, I would have one single database of consumers. Why? Because Europeans travel too much. They go here and there to many different countries. So this rationale, it's not local, it's more global. The other aspect tends to be more local. And being a local company or being a local business actually, it demands you, in my opinion, to have most of your platform also local, to have the flexibility to react to what consumer wants to what competition is doing and so on and so forth.

Okay. Fantastic. Let's move on. We've got a couple of write-ins and a couple more people on the call. So we just have a few minutes left. So we have one from Scott at [ Tencor ] saying, thanks for doing this. So thanks for doing this, Diego. Can you give me a sense for what the Clube penetration is among monthly active users? And what is the difference in frequency among those customers versus the average?

Yes. So the penetration, it's around 30% in terms of users, while Clube orders is 40%. The frequency generally starts at 2, 3x when the client enters in the program, he is operating at 2, 3x a month, and he goes to around 5, 6x. So that's the change. Of course, you cannot just do the math because there is maturity of the cohorts. But if you get the older cohorts, meaning around 18 months that's the uplift that we saw when we had only food delivery. What we expect now it to reach something even higher because of the new adjacencies that enter at Clube.

Okay. Great. A follow-up from Scott on that, which I think is also interesting. Can you discuss what percentage of Clube's orders are coming through a user searching for a specific item versus Clube coming to them with the offer?

Yes. So I don't have the number here, but I can tell you directionally, most of the orders are done because they enter at Clube domain, let me call this, which is the section where Clube is, and they choose there. So which means that more and more users are already getting use it to look for the result -- looking for the benefits that Clube provides.

Helpful. Let's go back to the phone and [ Matty ] from HSBC.

Just one question actually on -- so I just wonder about regulatory aspect of what you're doing. I mean loyalty program like this makes you even stronger in already very strong market for you. So is there any regulatory concern or aspect you are thinking about, you are preparing for? I mean I wonder whether you may attract some kind of antitrust because you are dominating the platform or ecosystem space as well? Just some color around that.

Yes. That's an important point. I don't think we are dominating. I think we are leading. And why? Because of several aspects. First of all, there are many segments of this market that not only the app search. In Brazil, 30% of the delivery happens because somebody called using it's phone, around 30% is through apps and around 30% it's through WhatsApp. So we don't work on the call -- on the phone segment. So what I'm trying to say here is that we don't dominate -- we can be very strong in one segment, but not necessarily dominate the market because there are much more market than where we operate. So that's the first thing. The second thing is we already have a discussion with the Brazilian antitrust body. It happened 3 years ago, and the result was an agreement made by ourselves with them in order to comply with some restrictions. So they didn't apply any penalty to us, and they didn't issue a sentence saying that we were doing something wrong. It's not about this. What they said is, I just need to understand better if your power, if your forces can drive to a negative aspect to the market. And we signed a deal with them. And I'm going to give you an example of part of this. We can't be an exclusive partner of key accounts, such as McDonald's, Burger King, Outback and so on and so forth. 3 years after, we're still in the same strong position. Why? Because that's not the point. We are leaders here because we are innovators. We are entrepreneurs. We create. We change. We have a loved brand. We have a great customer support. People love to be here. So that's the thesis we are telling to the antitrust body. And 3 years after the signing of the agreement, that's what I believe it's getting even clear to everybody. So something can happen in the future? Of course. But there is something at this point in time that worries me? No. Is there any talks to the antitrust body? No. So at this point, everything is clear for us to speed up the growth process here.

Just a quick follow-up because you mentioned something very interesting that of the overall delivery market, you're probably only doing 30% of the market, right? So you still have almost 60%, 70% of the market you are not participating. Is that an opportunity at all, do you think?

Yes, yes. So if you go back to 2 years ago, this 1/3 of the market that was on WhatsApp were like blue ocean. We entered there. And currently, we are the leaders of this market. We are the leader by far of the WhatsApp ordering. When we look to the phones, I don't think it's an opportunity in terms of having a product there, but it's an opportunity to more and more change people behavior. So there is a part of the population there that doesn't use too much smartphones, apps and et cetera. And this part will be tougher to bring here. But there's other part that use. So what will happen is this part will be smaller and will come to these other two segments. And the other part will stay there for a while.

And by the way, it's a fantastic product placement, Diego, with that mug. Let's go to Robert from Kepler, please.

And my first question is, I would like to understand a bit more about using Clube to grow the adjacencies and then the impact on the customer acquisition costs. Clearly, you see significant synergies for your food delivery loyalty program with the adjacencies that you're pursuing. Can you maybe quantify, give us a feeling how meaningfully you think leveraging the Clube program to cross-sell can bring down the customer acquisition cost to scale adjacencies like fintech? And then maybe my second question is about logistics. Yes, of course, the delivery of food from restaurants to people's houses is an interesting mathematical challenge with regards to the optimization of routing, stacking. And of course, your business has quite some scale. So any efficiencies with regards to logistics can yield quite some results. I would like to understand whether you think that you have just scrapped the surface of optimizing logistics? Or do you think that future algorithms, technologies dedicated to optimizing logistics can yield more benefits?

Yes. So first of all, Clube, it's great lever to grow the other adjacencies with a lower CAC or CPO, cost per order. And the reason is simple. The reason is, first of all, there are no big platforms in Brazil working on these adjacencies. They are not strong. They are simple. Most of them are digital platform from the incumbents, and they don't play technology and product in the same level that we can do. So that's the first thing. The second thing is there is a clear message from the consumers that they want something much, much, much simple to do their shop, to buy a cosmetics, to buy a medicine and so on and so forth. So more and more people are just raising their hand and saying, I don't want to go to the store and I want something very simple and so on and so forth. The third thing is for this business to work well with the competitors, they have to have a very efficient logistics, which goes back to your second question, but I will not answer right now. And it's tough to have an efficient logistics when you don't have scale. That's why when you go to 2016, '17, '18, everybody was burning like hell because there was not a density to allow optimization of cost. So with that in mind, what I can tell you is there is demand. There is a clear case for the consumers, which is iFood, and there is a great product here. What is the end of this rationale? The industry and the merchants, pharmacies, pet shops and bars and et cetera, they will jump at Clube. Actually, they are already jumping. Out of the top 10 drug store chains in Brazil, 9 is already investing at Clube. There's a 4 months program for them. 9 out of 10. Why? Because the demand is there, the product is there, people want to do something like this. So I clearly can see an expansion of these categories with a lower CAC and CPO. What does it mean? It means I will grow faster in these adjacencies with a better unit economics. That's what you should see in the numbers, right? And theoretically, you should grow faster with a worse CPA and CPO because you need to expand type of clients that we're bringing in. So the numbers that I expect is this as a consequence of what I told you. And we should see this already this year. The second point regarding logistics is our logistics is getting more efficient for two reasons. The first one is density. The more we grow, the more we can have our models with a better output in terms of optimization while increasing the gains of the drivers. That's one thing. The second thing is we are testing for the last 12 months, different business models of logistics, which I'm not going to go into details here. It's not necessary, but it's different from the one that we have. When I say different business models means the way we price, the way we organize the demand and the supply, the way we do demand shaping and so on and so forth. The results here at this point in time are good. If we are able to expand to roll out these tests, we are doing four cities to the remaining cities of Brazil, we're going to see another phase of optimization. Am I saying this will happen? I said we need to end the tests. I'm confident, but we need to end the test. So I expect to see a new phase of optimization in terms of logistic costs if we are successful with these tests, which I expect to finish more 3 months.

Okay, Diego. We're almost there now. You have room for two more. Do you think you can do it? Okay. Here's one. Can you tell me how important is it at being first to market for food delivery success? Once you're a leader, is it difficult to have that share you served? So that's the question.

It's irrelevant. If you have $1 billion -- if you give me $1 billion and let me build a product versus a company that has a terrible product, I will win. I will. Of course, again, if there's a clear leader, you need money to attract the consumers and to say, hey, I'm here, come, there's a new option here much better than the one that you are using, right? So being a clear leader doesn't give you any advantage except one, density in logistics, which is a great lever. But in the end -- see, this is like a very tough business in terms of customer service, very tough business. And when I say customer, I'm talking about consumer, I'm talking about restaurants and drivers. It's tough. It's very tough. I mean you buy a book, you don't get your book today, nothing happens in your house. You buy a meal, it doesn't get delivered. You get mad. Your kids need to go to school. They need to eat. You need to go to a conference call. You didn't have lunch. It's difficult. So having a great product here, it's mandatory. It's mandatory. That's why we focus so much until 2021 when everybody in the world was saying, become a super app. Being a super app is the obvious path to everybody. And we didn't do that. Why? Because I want to be the best. Let me give you one example of it. My customer support currently treats 80% of the incoming calls and chats through an agent that we developed internally. This agent performs better than humans. The NPS, the CSAT, which measures the satisfaction of the consumers are better with the agent versus humans. You know what does it gives me? Food delivery works like this, lunch, dinner. If I don't have this kind of solutions, I will never be great. You need to be great. So a leader can be beaten if he's not great.

Wow, it's fantastic. So I guess product quality determines the depth of your moat.

Yes.

All right, my friend. I think we have to leave it there. I want to thank you very much for your time. I think that was really interesting, and I think we're going to have a lot of follow-ups to this. But thank you very much for your time, and thank you all, everyone, for joining. I think, Luke, we can close down the call now.

Thank you very much. This concludes today's call. Thank you, everyone, for joining. You may now disconnect. Thank you.