Méliuz S.A. (CASH3) Earnings Call Transcript

So good morning, everyone. I am Marcio Penna, IRO from Meliuz. It is a pleasure to be here once again to share with you our numbers relative to Q2 '26, yet another great quarter with many records. We grew EBITDA 38%. We have reached BRL 109.6 million EBITDA last 12 months and BRL 21.3 million in cash this quarter alone, among other indicators that you'll see throughout the presentation. Last night, we announced an important material fact about a new buyback program which will also be addressed today during my presentation. So this conference has simultaneous translation into English. If you wish to listen to it in English, just click on the interpretation in the bottom part of the screen. This event is also being recorded and will be made available at our IR website. The respective materials can already be found at our IR website and also have been filed at CVM. At the end of the presentation, we'll start a Q&A session. [Operator Instructions] I'd like to turn it over to Gabriel Loures, our CEO, to kick off the presentation. Over to you, Gabriel.

Thank you, Marcio. Good morning, everyone. As Marcio said, very happy to be here with you once again, yet another strong quarter both in terms of operational results and also in terms of new things, material facts, which was announced yesterday about the canceling of shares for the first buyback program, which we opened October last year. We closed this program in record time, less than a year, 9 to 10 months to be exact. We did it exclusively with the cash generated by the operation, and we started a new buyback program. So I will start with our operating results, the growth of our results, our verticals. And then I'll be talking about our new buyback program and our treasury strategy as well, which was also announced yesterday. Okay. Just to start, a bit of our large numbers. We are quite happy because we proved once again that our operating results are not only accounting relative. They are also cash generation. In the first half of the year, I think the most important result is the generation of BRL 21.3 million in cash in Q2. Last 12 months, it's over BRL 41 million in the operation, reinforcing that our EBITDA, our accounting results are not only accounting based. We have no debt, no high depreciation, amortization. Everything we generate is translating into cash, and that's yet another proof of that. Looking at the operating numbers and accounting numbers as well, our net revenue consolidated terms grew 18% vis-a-vis Q2 last year, BRL 115 million in the quarter. And in the LTM BRL 500 million in revenue, a growth of 26% vis-a-vis LTM for last quarter -- last year's second quarter, Shopping growing significantly. We're going to be talking about that in a moment. Shopping grew 30% year-on-year. And for the last 12 months, we grew 44%. So very strong growth results, but more importantly, without losing margin. So adjusted EBITDA grew almost 40% year-on-year, BRL 17.3 million. Once again, the only adjustment here is the Bitcoin, which is a non-cash effect. The rest is a de facto result from the operation. And in terms of the LTM, BRL 110 million in adjusted EBITDA, yet another historical record, growing EBITDA quarter-on-quarter. And every quarter, we present a stronger quarter when compared to previous ones. In terms of new things, we had the cancellation of the shares. So we canceled about 8% of our total equity, 10% of the free float. So the share of all of our shareholders account for 8.5% more. So more participation in EBITDA, net profit and also Bitcoin. So the Bitcoin yield is the same metric, 8.4%, okay? So strong results once again. And I'll be going into more detail as we move forward, both in terms of operation and in terms of products that we have. But I also have to start the presentation by commenting on our strategic plan, which we announced late last year, which hasn't changed. We have 4 main pillars that you can see on the slide for the operating front. And our main objective is to grow return to our shareholders. And I'll go -- I'll walk you through all the slides, all the pillars. We are comparing what we promised last year and what we are now delivering quarter-on-quarter across all the verticals we have in our company. Looking at the growth in top line with a focus on Shopping. When I look at a broader period for the past 3 years, you can see that our total accounts have grown at a pace, which is impressive, increasingly higher. Second quarter '23 to second quarter '24, 5 million accounts were added and then 10 million accounts '24 to '25. And now once again, we have added 11 million of accounts. So every quarter and every year, we add more users to our database. And we have more users actually accessing our platform. On the right-hand side, the number of new buyers. That number is not officially announced, but it's important to notice that on the right-hand side, that it's not an accumulated number, no. This is -- otherwise, the growth would be a lot more than that. We are talking here about the growth of new buyers. So first-time users accessing our platform, and that number only grows. Year-on-year, we add new buyers when you compare year-on-year, '23, '24, and '25 and '26. So that is a record. And that's a very relevant result. We show that the base growth that we talked about last year, we were among the 7 most downloaded apps in Brazil, and that reflects in the number of new users, as you can see, they're using the app, using the site. And as a consequence, we increased our revenue. So an increase in the number of accounts, of course, helps us increase our active base and reflects in the revenue. So as important as the growth in new buyers is to reinforce the following. We have to retain them, right? And so here, once again, this is the secret of our success, the retention level throughout time. We bring increasingly more users and those users are buying the columns on those graphs and those users are retained in the long run. This is the number I showed last quarter, a long-term view. We show this once a year, and we show that retention of revenue of our users for the last 10 years is 219%. In other words, whatever we acquired back in 2015, '16 is today generating more revenue than they did back 10 years ago. So we have excellent retention numbers. I said this before, and I'll repeat it. I do not know of any other B2C platform with such a high long-term retention. I'm not talking about users retained for 2 months. We're talking about 5, 6, 10 years. That's the value of our platform. And that's why growing our user base, as we've done for the past few years, helps us to create value in the long run. And that's what we also see on the next slide when we show our revenue. When we compare Q2 '25 to Q2 '26, we grew 18% total net revenue. And when we look at the LTM vision for the last 12 months, that growth was even higher, 26%. In practice, that result is a consequence of the growth in our Shopping revenue by 44%. So our core business, the business which has been here since 2010, 2011, it's growing 44% in the LTM metric, which is quite strong. This was offset by a drop in financial services. We have already announced the termination of our partnership with the bank. So this will lead to a drop in revenue along that vertical for 26%. And we also had a negative result for other companies, especially because of the Picodi. And when you add all that together, we had a growth of -- a drop of 5% actually. When you add it all, we grew 26% of revenue in the last 12 months, growth healthy, which was healthy and strong without forgetting the margin. And of course, focusing on our core business which is the Shopping vertical. A bit of a breakdown of those numbers. I had the same chart last quarter. This is a chart that shows the growth of net revenue for Shopping Brazil broken down across its different divisions, e-commerce, our traditional business with commission with our retail partners online and beyond e-commerce, the new verticals we launched in the past few years. Here, we have Nota Fiscal, research, games, several other business verticals that make up our beyond e-commerce front. When you look at the Shopping Brazil net revenue, it grew 32% quarter-on-quarter and 44% LTM. And the breakdown is 19% of e-commerce, traditional business, 20% growing year-on-year. The beyond e-commerce more than doubled at 114%. And as a consequence, we have increased its share. A nice way to see it is the addition of absolute revenue across periods quarter-on-quarter. We have added in terms of revenue about BRL 22 million in revenue, BRL 22.2 million to be exact. And it's basically half and half. Half of that growth came from e-commerce at 30% a year growth and half of the growth came from beyond e-commerce. which we have doubled. So very happy with those results. And the trend is that we continue to see an increase in the share of beyond e-commerce within the whole business because its growth rate is higher. When you look at e-commerce specifically, I think some important takeaways here when you look at the year-on-year LTM, we grew by 26%. The e-commerce revenue in the same period, GMV grew by 5%. So there's a huge impact of our take rate growth, which moved from 6.6%, 6.8% to 7.6% earlier in the year, Black Friday, first quarter this year and now at 7.4%. I heard some comments from analysts. We have been talking to several analysts, of course, since the announcement of the results last night and some have asked me about that percentage, the take rate percentage. Why did we see a drop quarter-on-quarter, first to second quarter? I'd like to reinforce that there is a very important seasonality. The Q4 and Q1, they had a very strong impact coming from seasonal events such as Black Friday, which is only natural that we see this sort of variation of take rate across quarters. That's why it's good to look at it from the last 12 months view. Then you clearly see an increase of the take rate when you compare year-on-year, '24-'25, '25-'26, moving from a level of 6.8% or 7.1%, 7.2% to a level of 7.4%, 7.6%. So that's a quite positive result at the end of the day. And when you look quarter-on-quarter, it's a very stable trend. If you look at net take rate, which shows it more explicitly, you see that all the historical series we have a very flat behavior, 6.2%, 6.3%, which we consider to be a healthy level and which we always pursue going forward. Moving away from e-commerce and going to beyond e-commerce, I have this metrics, which I saw last quarter. Just a quick recap before I move forward. The role that products play within our ecosystem, there are products which have a use frequency, which is higher, but an LTV per user, which is lower, whereas e-commerce has a higher LTV, higher tickets, higher commissions. But on the other hand, the frequency is lower when compared to other products, considering Nota Fiscal, research surveys, those are higher frequency products. Our main objective here is to have multiproduct users because I have high-frequency users and combine them with higher ticket. So I have higher LTV users and combine that with higher frequency. So I have the best of both worlds as I increase retention. So those 2 bits of our ecosystem, they are very synergistic. That's why we break it down like this, beyond e-commerce and e-commerce because I have very different factors. And the great piece of news we had in Q2 was our ability to increase that and lead users to adopt or to access other products. So the cross-sell has really increased. So users who were using beyond e-commerce now use e-commerce. That number grew by 87% in one single quarter. So we had 100 users at one business level migrating to the other. Now we have 187. The impact is not immediate. We have the different seasons that -- which is our e-commerce business as usual. But it shows that in the long run, we'll be able to continue to grow our multiproduct base, be it a user from e-commerce using e-commerce or beyond e-commerce users moving to e-commerce. And an important part of that number, that result comes from intelligence. I have an AI section, and I'll talk about that in a moment, but it's nice to show a bit of those changes. We have generated over 80 million recommendation average per month based on the consumption behavior of users on beyond the commerce offline. So that's somebody who goes to supermarket buys diapers. And then they also buy those diapers at e-commerce because perhaps it's cheaper, quicker freight or free freight. So we have managed to do that very efficiently, and we want this to grow. There is room for that to grow across products and across platforms. We have just started that journey along recommendations. And then talking about the 2 products, in addition to growing the user base, which was the focus of our first vertical, the second was to grow revenue per user. And the mission I have shared with the market and we convey to our team has been met and really well at that. This is the chart of our revenue per user, considering active users and direct revenues. When you look at ads, it's difficult to have a direct correlation. So I'm talking about for each user based on what they actually did within our ecosystem. When you look at e-commerce, that number grew 37% LTM year-on-year. And in the e-commerce, that number grew by 78% year-on-year also in the other LTM view. You can see by the chart that it is a very significant growth when you compare it to Q1. So we continue to grow revenue per user across both verticals. As I mentioned before, a totally different LTV when you compare both verticals. So -- of course, the ARPU, the per user revenue is different, a different base across those 2 users. But the idea is to show the growth, the growth of both verticals which are happening. We retain more users, they become more frequent, they use more verticals, and then we expand the share of wallet across the board. And once again, I'd like to show this number under LTM view because we do have seasonality. If I was focusing on Q4 '25, the number would be higher, of course. The revenue would be higher because of seasonality, because of Black Friday, specifically. So the LTM view shows that we are having an average across the 12 months, including the most recent quarter and the most profitable quarter, so that we are growing anyway across all fronts. Well succeeded from -- and we started the turnaround process back then. We want to grow, but also maintaining or preserving margins. That's what we're doing. Our revenue moved from BRL 393 million LTM to BRL 495 million. So we had added BRL 100 million, that's a lot. We grew fast, especially given the already relevant level in 2025. And our percentage of cost on net revenue remains quite stable. When you look at the cash back, it moved. Yes, I think he crashed -- he's going to rephrase. Can you hear me now? Yes, yes. I think we can, yes. So let me talk a little bit. I said our revenues were growing. We added BRL 100 million in revenue, BRL 393 million to BRL 495 million. And our level of expenses remained flat at 46% in Q2 '25. Expenses with cash back, that number dropped 2 percentage points. We are now a bit more efficient at 45%. Fixed expenses, suppliers, personnel, AI, 29% down to 25%. So here yet, we have also decreased by 2 percentage points and our marketing expenses moving from 12% to 10%. That's a vertical that we intend to increase actually in terms of investment in marketing. We do not see a payback. That's why we are quite cautious in terms of investments in marketing, but our aim here is to increase that a little bit. Everything I want to reduce in fixed expense, I want to offset in our marketing investments so that we can continue to grow while preserving margin. Unlike other quarters, I'd like to give you a breakdown of our cash back costs. When we talk with analysts, we saw some -- we received some questions about the cash back as a percentage of revenue, which in Q2 '26 reached 58%. So cash back as a percentage of net revenue for Shopping hit 58% in Q2 2026. It's higher than Q1, which was sitting at 44%. And some analysts ask, why did it grow so much, 44% to 58%? If you look at Meliuz track record throughout time, there are seasonalities involved. Q1, we know has revenue coming from Q4. We had GMV for Black Friday and so on and so forth. So it's only natural that you have a different sort of behavior. But when you look across all quarters, a longer historical series, we are -- it's actually below our historical average, which is about 59%, 58.5%. Some quarters a little above, some quarters a little below, but always ensuring that consistency. So nothing has changed. We are not delivering more cash back than before. The rationale here is to ensure sustainability for the business. Cashback is the most important tool for us to guarantee retention. We do not intend to reduce that number in a very aggressive manner. In practice, we are maintaining those costs at a healthy level. So when you compare Q2 '26, Q2 '25, we were more efficient, 4 percentage points. We moved from 62% to 58% in Q2 '26. Not for nothing. In an LTM view that percentage of cash back per revenue has dropped a few points because the current quarter was more efficient than the same quarter of last year. That's always the correct comparison. Compare the same quarters, the same seasonality or looking at the LTM, including more periods to reduce the impact of seasonality. The consequence of that very strict cost control appears in the adjusted EBITDA comparing quarter-on-quarter once again and LTM. We never compare 1 quarter with the previous quarter. So quarter-on-quarter, 40% growth, half of the year -- half of the year, 60% growth year-on-year, 74%. We're being very efficient, generating a much stronger EBITDA quarter-on-quarter. Our EBITDA margin from Q2 was 16% vis-a-vis 12.8% in the previous quarter. When you look at the whole half of the year, 15.3% vis-a-vis 15.1% and LTM 22.1% vis-a-vis 16.1%. So across all comparison basis, we are growing EBITDA and bringing more results. And that, of course, reflects throughout time. We moved from BRL 38 million EBITDA last 12 months to BRL 109 million. Just do the math, right? We practically doubled -- more than doubled EBITDA, moving from BRL 40 million to BRL 110 million, almost tripling EBITDA in 2 years, and we moved from an EBITDA margin back in the day at 11.3% to a margin of 22%, have doubled EBITDA margin and tripled the EBITDA in this period shown on the chart, 24% to 26%. On an annual basis, we grew 74%, the absolute EBITDA and added 5 percentage points in margin. So it's a very strong momentum, and we continue on track as we deliver results, both from revenue growth or from EBITDA growth. Of course, we can always improve, but this has been, without a doubt, a very strong quarter. To reinforce this perception, the variation of our cash. I'd like to reinforce this. Our EBITDA translates into cash for the business. We moved from EBITDA accounting 14.8% negative. There is an impact of Bitcoin, which is a non-cash effect in Q2 which will be partially reversed based on the current valuation. Of course, it depends on future numbers. But in a hypothetical scenario, if we look at the numbers today, there would be a reversal within this amount. And the adjusted EBITDA for Q2 was BRL 17.4 million. This is a number that appears on other pages. With BRL 17.4 million of adjusted EBITDA and ex-Bitcoin generated BRL 21.3 million in cash. So there is no non-cash impact. Of course, there are seasonal effects, but what we can clearly see is that our conversion from EBITDA to cash is quite strong, and we're able to generate BRL 21.3 million in cash in Q2. When I look at the last 12 months, we generated more than BRL 40 million in cash, something close to BRL 41 million, BRL 42 million in cash for the last 12 months up to the second quarter of 2026. And why is this important? Because cash generation is the main tool for us to generate our Bitcoin and increase our participation of our shareholders in our buyback program, which I'll be addressing in a moment. But before, I'd like to talk about the fourth block of our operational strategy, which is the AI front; how are we using AI to leverage our results. This is a slightly different way to approach it. This is a time line approach where we started, where we were and what -- how have we evolved in Q2 and where we want to get to. So we have split engineering products, analytics, the use of data across the company and the business and support areas, including operations, service, commercial growth, all the business areas under the umbrella of the company. When you look at the fourth quarter of 2025 and even before Q2, Q3 '25, AI was a support tool for most companies, I would dare say, sort of a chat box helping us -- help us writing right, but not very much integrated with other processes. We turned the year and we decided we had to change. We are lagging behind. We need to speed up. So we accelerated the process early this year and the results that I showed in our last call was quite evident -- were quite evident. The first few we using -- moving AI from a support tool to an integrated tool 90% of our codes are written by agents. So we use more AI. We created the whole basic infrastructure to integrate our database to create analysis scenarios, take rate, GMV adjustments, creating skills for AI to be able to read and interpret and make the analysis. And we created several simple automation. So co-creation of content, CRM, ads, several processes which were automated, registration of cash backs, things which are manual, but simple. Processes that consume time, or time consuming, and we're able to optimize and free those people up to do other things. In the second quarter, we have changed even more. When I look at engineering products, we talked about the end-to-end flow. The product is not restricted to code writing. We start translating a business need into a requisite to translate into a code. And it ends adds when that product reaches the final user. In the first Q, we focus on the middle part, which is to actually to code. And then in the second quarter, we improved the flow end to end. I mentioned that in our letter. Our engineering team, designing team, they are automating all those flows now. And we are able to generate a business need, translate that automatically into ERPs, which are read by AI. And throughout the development of all that code, we are already doing that using AI, which is also testing that. We are rolling that out to all teams. And the idea is to make that flow increasingly more automatically. So our teams will not only coding, but they will be reviewing, editing, ensuring recommendations are well done and that AI is truly translating what has been specified early on in the process. Looking at analytics as a very important part, which is a real-time insights, how can we make sure we move from the ad hoc analysis as we had before in Q1? How can I integrate all that with the database? And now I have to have real-time insights. How do I know the performance of a partner in the past 30 minutes has dropped and that might create a problem. How do I know that test? I uploaded our partner landing page is losing 10% it is time to turn it off. How can I make sure that looking at our partners, they can see real time the results of their campaigns? So we have moved forward significantly on that front. We are doing really well in automating those real-time insights vis-a-vis ad hoc insights. And we have an increasingly more automated process. So we already managed platforms, TikTok and other, can analyze performance much faster. We have a very more important back office to do things more automatically, faster, more expeditedly using AI agents. And of course, customer service has become more effective, generates more satisfaction on the side of the customers because we are using data intelligence from end agents to better service. That's what we've done so far. The results were very nice in Q2, but there's a long way to go. We are just in our first steps in this AI journey, even though we have advanced significantly. What are the next steps then? When I look at engineering and products, we want to have a 24/7 product. So it will close the day, go to bed and the agents will continue to work. So next morning, they will be able to review the code and be able to upload that for production. 24/7. That's we are developed -- that's how long we are developing our products, and that increases our development speed and the speed between somebody thinking about an idea and the final user actually putting that to the test. In analytics, we moved from real-time insights to autonomous decisions. So instead of knowing that the A/B test is 10% worse than I need someone to go there and address a problem, I want the agents to make decisions on their own. And the same thing goes for business and support, increasingly an autonomous process. Those are the next steps, a lot to go. Now we already have 46 days since the end of the quarter. We have advanced significantly across many of those topics, and I'll share that in our next calls. At the end of the day, AI is not a target. I always say that we cannot make a mistake. Our goal is to generate EBITDA, cash return that's not using AI. AI is a way to get there, and we need to use AI to be able to improve those results. That's a game changer, but not the goal. Looking at the clear results we had, the first metric is the number of PRs per engineer. That number is already 160% higher than it was in Q4 '25, 33% more than the previous quarter. So we have increased our speed by 33% quarter-on-quarter and our annual revenue per employee has already grown as well. When you analyze the fourth quarter to this quarter, 20% growth and quarter-on-quarter the growth was 10%. We are now more efficient and generating more value for our users, partners and to our whole ecosystem. That was the main takeaway I'd like to share with you in terms of operating points. And then I'll move on to our buyback program we announced yesterday and what are the next steps for that. And the idea is the same to generate value to our shareholders and grow our numbers. I'd like to start by recapping our buyback program. We announced that in October 8, 2025. That material fact we wrote. If you take our market cap minus our net assets minus Bitcoin, the number is BRL 10 million. That's back in October '25. On the right-hand side, we have the last 12 months results. 12 months later after that material fact, the last announced number was in the second half -- the second quarter, then we had an EV of BRL 10 million and generated in 12 months, BRL 110 million in EBITDA, BRL 41 million in cash. And with those BRL 40 million in cash we generated, we rebought 37 million shares. In other words, we generated 4x more cash than it was our enterprise value back then. It was quite mismatch. And we did our role to capture that asymmetry. We bought -- in 12 months, we bought 8.1% of our total capital, 10% of the free float. That's the limit we had. So 9.1 million shares. And when we announced yesterday, when we announced the material fact, we had -- we canceled 100% of those rebought shares. So now we have 8.1 million fewer shares in circulation, which means that the share of all our shareholders from now on is 8.5% more. So looking at the asymmetry of EV of BRL 10 million back then, we captured that mismatch. And with that, we generated value to our shareholders, increasing the share of all those shareholders in 8.5%. Of course, that results in a Bitcoin yield of 8.5%. So they now have 8.5% more of Bitcoin in the operation and also 8.5% more of all of the business, EBITDA revenue and other results that we have achieved. In an annualized view, the result of the program was 11.5%, which is also quite relevant when you look at the Bitcoin yield metric. So in short, we have delivered an increase of 8.5% of Bitcoin yield to our shareholders because of this buyback program, capturing that mismatch. But that scenario continues to exist. We have a market cap closer yesterday numbers before the cancellation of the shares. Of course, this will change once those shares are canceled. But as of yesterday, August 4, 2 days before, our market cap was BRL 509 million. Removing cash, the treasury shares and the Bitcoin math back then, we reached an adjusted enterprise value of around BRL 200 million. On the right-hand side, our adjusted EBITDA for the last 12 months is BRL 109 million, which is growing 74% a year. So that gives an EV EBITDA of 8.1x for a company that's growing EBITDA at 75% rate and growing revenue at 26%. We continue to have very compressed multiples when compared to our peers, and we continue to be the peer that most grows revenue and EBITDA. So the asymmetry I mentioned just now, which we captured in October, it's still present. So there's another way to look at it, looking at the comparables. Here, we plotted on this chart all the Ibovespa companies which announced results for Q2. The ones we have not announced, we're using Q1 numbers. But still, we remain the company that has the highest growth rate in terms of revenue, and we continue to be the company with the lowest EV multiple on net profit. And I use net profit because it helps us compare with -- compare companies with debt with those with no debt. We are negotiating at 1.8x EBITDA. We have no debt. And consequently, we are negotiating at 3x net profit and growing really fast. So based on that information, we decided to create and launch a new buyback program, which we announced yesterday. With the new buyback program, we will buy up to 8.1 million shares, which means 10% of our free float at about 8% of our total capital. And then '26, all the ones I mentioned throughout the presentation. Adjusted EBITDA of BRL 110 million, growing 74% a year, BRL 40 million -- more than BRL 40 million in terms of cash generation for the last 12 months, BRL 260 million in net assets, no debt and an EV EBITDA of 1.8x. That justifies this new buyback program. And to speed up the success of that program, we have also changed our treasury policy, and we are now looking at our net resources like this. We have the cash generation from our core business as part of our net resources for cash. And we have our Bitcoin treasury, which is also a net asset. We will assess and manage that in an integrated fashion throughout time, and we will maintain the same objective, which is generate Bitcoin yield and value to shareholders. The final target doesn't change, but we'll do that using all our resources across 2 verticals to acquire more Bitcoin through cash generation or via derivative operations, and we will also use the resources to rebuy shares. So the last buyback program was done with our operational cash balance, can now be done with our Bitcoin balance. But to reinforce, nothing changes in our final objective. You want to generate value to shareholders and also generate Bitcoin yield. So nothing we'll do will harm Bitcoin yield we have to generate to our shareholders. The idea is to be more flexible and allocate capital in a more efficient integrated manner. And that's why we are launching this new buyback program. At the end of the day, this program and everything I've mentioned throughout the presentation is being implemented now, and we do that because we do believe in our historical results and the consistency of those results quarter-on-quarter but because we also believe we will once again be the largest loyalty program in Brazil, the best place for users to buy and win, the best place for partners to generate AI. And for us, we want to be the best company, the best program for investors to have results in results, Bitcoin yield and capitalization. Thank you so much. Back to you, Marcio.

Very good, Gabriel. Our first question comes from Ricardo Buchpiguel from BTG.

Your comments were quite clear about the seasonality, not only on GMV, but on revenue, on net take rates and the importance to use at a longer period. I'd like to confirm if it makes sense for us to imagine an increase in take rate in the coming quarters and also explore a bit how you see competition this year, other cashback platforms and also other marketing options for partners? And the second question that I have, I'd like to better understand the rationale around the Bitcoin sale. Do you think you will need to do that for the second program? Because when you look at the company cash position at BRL 75 million, you are generating cash. So if you could use cash generation to buy back, that could also be a question moving forward. When you look at future potential buybacks if you need to be more flexible.

So let me answer your first question, and then Marcio will address the second one. As for partners and peers and competition, of course, several e-commerce players have grown strongly in MercadoLibre, Shopee, Amazon, of course. And what we've seen from last year was an increase in competitiveness in the market. The Brazilian market has reached a moment where it's highly concentrated around MercadoLibre. It continues to grow strongly, but there are other players coming, Shopee growing fast, all the Chinese players actually growing fast in addition to Brazilian retailers who have maintained their focus on sales. Of course, that scenario is evolving, but we see e-commerce with very high potential, e-commerce growing throughout time, and we see that fragmentation in the market helping us generate value. We are a partner of everyone in the Brazilian e-commerce. So for us, we are helping our partners, some we have more, some we have less, but we still generate value for the market as a whole. As for the net take, yes, you are right about the LTM view. I made it a point to reinforce that. Of course, we have variations which are due to seasonality. So it's always best to look at the longer run. The net take rate, you can expect to continue to maintain our historical levels. We do not expect any significant change in net take rate going forward. Of course, short-term variations may happen up or down, but the long-term trend continues. Marcio, as for the report, the buyback program?

It's a great question, Ricardo. Despite our cash generation being above BRL 20 million in the quarter and the LTM over BRL 40 million, part of that cash generation is being invested in other things other than Bitcoin. So our idea is to use an asset, which is totally net of Bitcoin to, in a timely manner, be able to buy shares, use those resources to buy shares. So that gives us flexibility and gives us agility when the administration deems it the right moment to rebuy shares. So we can use that Bitcoin balance to do that when we see fit. But you are correct, cash generation in itself and of itself would be enough for a potential buyback program. But part of it is committed, of course, compromising with our investment, suppliers. So the Bitcoin, because it's net, provides us with that flexibility. That's basically it.

If I could complement, Marcio, the core -- the buzzword here would be flexibility and efficiency. We want to be flexible to conduct the buyback program, and we want to be efficient in doing that. But nothing changes in the major run target. Nothing changes in our goals, the targets we presented in terms of Bitcoin yield. What changes is that we're going to be looking at the cash in a more integrated manner so that we become more flexible and more efficient.

Our next question comes from [ Carlos Ereira ] from Conqueror Investments.

I have 2 questions. The first one is about this reclassification of expenses, other operating expenses actually that led that other revenues with the BRL 100.3 million vis-a-vis last year, which was 1.3% or something. I'm not getting the numbers right. Anyway, I'd like to understand how that rectification happened, right? If you could explain what would be the run rate of that line, other expenses, if you will, moving forward. That's one question. And within the same line, as for fixed expenses, they grew by 11% year-on-year. And when we look at it vis-a-vis the revenue, 27%, give or take, what kind of level were you expecting to have those fixed expenses vis-a-vis the revenue in the long run?

Okay. Marcio, would you like to start?

Okay. I can start, yes. Carlos, thank you for the question. Basically, we are talking about products that have generated revenue in the last quarters, which have increased the impact on the final bottom line. When you look at the second quarter 2025, many of the products that made up the revenue of beyond e-commerce, they were falling under other revenues and expenses. And why was that? Because those were less significant costs. We made that accounting change, that reclassification, as you mentioned, late last year. And from then on with a higher impact and a higher share of those products, we created the beyond e-commerce class within Shopping Brazil revenue line. That's why there is a difference between the Q2 number '25 and Q2 '26. In Q2 '25, there was a revenue from other products that are now no longer present because they do have a specific line within Shopping Brazil revenue. Of course, those are accounting changes, which do not affect EBITDA. They are net 0. So in the previous period, there were expenses, now they are revenues, not affecting EBITDA. When you talk about run rate for other expenses and revenues, given that we made this accounting change, it would be more normal to expect run rates are more related to Q2 '26, not Q2 '25. But it's worth remembering that other expenses and revenues have many variations, and they also vary across quarters. But it's normal to think that the run rate number will be more linked to Q2 '26 because of what I explained. Okay?

Carlos, you can also look at other expenses and revenues. From the moment we reclassified those entries, Q3 '25, so as of Q3 '25, that account, other expenses and revenues, it was already including only expenses, but no revenues because revenues had already been reclassified. But as Marcio said, this will vary quarter-on-quarter. So from second quarter '25, first quarter '26 rather, to second quarter '26, it dropped because we had a legal expense, which happened in the first quarter. So this will vary -- just to say, this will vary quarter-on-quarter. About your other question about fixed expenses, I think the best way to look at it is on a line-by-line basis. I have the personnel, software, third-party services and other expense and revenues that you mentioned. All of them are dropping in a year-on-year basis or moving sideways from 20% to 18%, software, 3.3%, 3.4%, sort of flat and third-party services also flat. So fixed expenses are called fixed not by chance. The idea is to stabilize those expenses. We don't try to avoid increases in those expenses even when revenues go up. So when you look at expenses over revenue, that number is going down. Those are the charts I showed during the presentation. So of course, we'll have variations. It's not endless, right? It's not an infinite chart. There will be a point where you need to reinforce your cost and platform personnel and so on. But we believe we are at a very healthy level now in terms of expenses over revenue. There is room for us to maintain that level in answer to both your questions.