--- title: "Mercado (Trans): Will develop an in-house e-comm Agent" description: "Below is Dolphin Research's transcript of the FY25 Q4 earnings call for $Mercadolibre(MELI.US). For our earnings take, see 'Latin America's Alibaba's Split Personality: Revenue Giant, Profit Dwarf'. I" type: "topic" locale: "en" url: "https://longbridge.com/en/topics/38871296.md" published_at: "2026-02-25T03:44:59.000Z" author: "[Dolphin Research](https://longbridge.com/en/news/dolphin.md)" --- # Mercado (Trans): Will develop an in-house e-comm Agent Below is Dolphin Research's transcript summary of $Mercadolibre(MELI.US) FY25 Q4 earnings call. For the earnings analysis, see '[Latin America’s 'Alibaba' with split personality: revenue giant, profit dwarf](https://longbridge.com/topics/38870288)'. **一、财报核心数据回顾** **二、财报电话会详细信息** **2.1 高层陈述核心信息** **1、电商业务** **Brazil strategy:** GMV accelerated +35% on strategic investments, notably lowering the free-shipping threshold. This lifted purchase frequency and attracted new buyers. **Scale effects:** Scale expansion translated directly into logistics efficiency. It demonstrated the company's ability to scale execution effectively. **Mexico:** The value proposition resonated similarly. GMV grew 35%. **2、AI 的具体应用与成效** **Ads:** AI-driven bidding algorithms and automated marketing tools improved seller ROI and raised adoption. Ad revenue accelerated +67%. **Acquiring:** In Brazil, AI helped identify high-value merchants faster, lifting per-merchant TPV and shortening payback. Acquiring TPV grew 25% in Brazil and 50% in Mexico. **Customer service:** Mercado Pago's AI assistant resolved 87% of interactions without human support. Millions of users now use it for card management, transfers and credit inquiries. **3、金融科技业务** **Positioning:** Mercado Pago leads on NPS across Brazil, Mexico, Argentina and Chile. It remains top-ranked on customer advocacy. **Preferred platform:** AUM grew +78%. This is strengthening its status as users' preferred platform. **4、财务表现** **Margin context:** Full-year OP rose 22%, but margins compressed. Management emphasized this reflects deliberate investments in the largest long-term growth opportunities, particularly fulfillment and card expansion. **Investment logic:** These investments aim to reinforce the ecosystem and deepen competitive moats. They are strategic in nature. **2026 outlook:** The company enters 2026 from a position of strength. All segments are growing fast, indicating current investments are bearing fruit and unlocking long-term value. MELI remains confident in sustained growth across markets where e-commerce and financial services penetration is still low. **2.2Q&A 问答** **Q:** You mentioned free shipping, 1P, cross-border and card investments together pressured margins by 5–6 percentage points. Given cross-border is newer than cards, where are you in that investment cycle? Which areas are at peak intensity and which may run longer? **A:** To clarify, the 5–6ppt range was to help quantify margin pressure from key investments. These include lowering the free-shipping threshold in Brazil, card investments in Brazil/Mexico/Argentina, 1P still on the path to profitability, cross-border expansion to China and the US, and continued spend in smaller countries. By stage: **in cross-border, domestic fulfillment is profitable, while intl fulfillment still needs scale**, **which will keep pressure elevated ahead**. 1P is profitable before overhead allocation, and scaling will help improve profitability. Cards are steadily improving, notably Brazil where older cohorts are already profitable. Overall, **all businesses are trending better, but faster-growing areas are driving mix shifts**. We are confident in these investments, which target long-term opportunities and improve UX. This quarter, e-commerce and fintech NPS hit record highs in Argentina, Brazil and Mexico. **Q:** Argentina's direct contribution margin fell QoQ. Last quarter had specific pressures such as repricing credit spreads, likely resolved, but logistics pressure may remain and Q4 is promotional. Can you break down the drivers and how to think about the margin into 2026? **A:** Argentina did see margin compression, though it remains our highest-margin market. Compression came from fulfillment, as we recently opened several new centers, raising costs YoY; card-related provisioning, given cards launched mid-last year; and higher financing costs YoY. While Q4 financing costs fell QoQ vs. Q3, they remain above last year. **Q:** On AI deployment, you noted demand/search-side and merchant/supply-side use cases. A long-running focus in agent-based commerce is ad monetization. Could ad dollars migrate upstream, and how are you preparing for that risk? It is **too early to conclude on revenue impacts from AI and agent commerce**. We **do not believe solving a single link such as search changes the game**; the key is delivering the best end-to-end experience. Search is one piece; reviews, on-time delivery, selection breadth, price, financing, fraud prevention and support also decide where buyers transact. We **focus on building our own agent experience within MercadoLibre**. We have first-party data to build highly personalized search and recommendation engines. If agent commerce scales, retail will shift online faster; we believe MercadoLibre remains the shopping destination of choice and can capture future revenue. Agent commerce outside MercadoLibre is also incremental. We already serve third parties via our ad tech stack, including Google Ad Manager, Disney, Roku, etc., given our unique data, customer understanding and attribution, which are hard to match. In short, we are confident in capturing traffic through our internal agent experience, and we also see agent commerce as an additional ad revenue opportunity. It may accelerate budget shifts from offline to digital, expanding the pie. We remain positive and focused on building the best ad stack and the best internal agent experience. **Q:** On agent solutions, your release says Pago's agent can resolve queries, but it seems more capable. How could the Pago agent, via increased interactions, influence borrowing, saving or adoption of new financial products? Also, what is your vision and timing for a personal agent on the marketplace? **A:** We are very excited about Mercado Pago's AI assistant. It **primarily resolves user questions and issues today**, and is already powerful enough to execute nearly anything you can do in Mercado Pago. For example, users can ask for all bills due next week, and the assistant will list and pay them. Currently, **the vast majority of user interactions are handled autonomously by the agent, without human intervention**. We **have not started using the agent for cross-sell yet, but we will**. In a conversational setting, the agent can present available credit or card offers. We believe the opportunity is large. The system will become more proactive, acting as a personal finance advisor to help with asset allocation or recommend suitable credit products. On the acquiring side of Mercado Pago, **we see opportunities to help merchants handle pre-/post-sales issues** and integrate with typical platforms, ultimately enabling payments using Mercado Pago credentials. We are excited about AI's potential in Mercado Pago. **On the marketplace, beyond AI-driven search and recommendation, the seller assistant is already live**. About 20% of GMV receives some form of assistant advice today. It has successfully helped sellers optimize listings, reduce time-to-list, improve reputation and handle parts of customer support. **Q:** Ads grew well this quarter and penetration is accelerating. Which products or geographies drove growth, or was there a specific accelerator? **A:** We are pleased with ad performance. On an FX-neutral basis, revenue accelerated +67%, with higher adoption and spend, mainly driven by improvements in our tech stack. Growth was broad-based with no single driver. We have products embedded across the value chain, including bidding, bid management and placement optimization, all supported by an improved, easy-to-use front end. Specifically, our budget orchestrator lets sellers reallocate unused budget across campaigns that were previously budget-constrained. We extended budget boosts for seasonal events, combining predictive signals to ensure sufficient budget in peak periods. We also launched performance-based recommendations and AI agents that support advisors and interact directly with sellers, accelerating ad penetration among mid- and long-tail sellers. In short, many tools are scaling, with AI used in several of them. The team has made great progress, but there is more to do. Ad revenue as a percentage of GMV remains small relative to potential, and we are happy with current results. **Q:** How do you balance sustaining 30%+ GMV growth in Brazil with margin pressures into 2026? Given share gains over recent quarters, should we expect further operating leverage in Brazil? **A:** **Margin should be viewed in the context of growth**. Most pressure comes from deliberate investment choices that are driving strong growth and improving UX. In Brazil, these investments have driven GMV growth and share gains, with total revenue up 45% YoY. NPS is at record highs, also an outcome of these investments. We **highlighted cross-border, 1P, lowering the free-shipping threshold, widening free-shipping coverage and better use of idle capacity**. We are comfortable with these investments and current margins because we are seeing growth, share capture and better UX and engagement. We have long said our focus is capturing the large opportunities in commerce, fintech and ads, investing decisively even with near-term margin pressure. We manage the biz for the long term, not to optimize near-term margins, and remain confident in the long-term margin trajectory. **Q:** On lowering the free-shipping threshold, how did performance track vs. expectations? Frequency seems to have risen; was that in line with your model? Also, Brazil's unit delivery cost fell 11% — is that sustainable as the economy network scales? **A:** We are very satisfied with Brazil's threshold reduction. This is our third reduction and the result mirrors prior rounds: growth accelerated, frequency rose, conversion and retention hit records, new buyers increased, NPS peaked, sellers and listings expanded, and share reached all-time highs. It matched our planning. Volume growth accelerated from +26% in Q2 to +45% in Q4, and GMV from +29% to +35%; given our scale, this is substantial. New buyers since the Jun policy change buy more categories and items and retain better vs. prior cohorts. On logistics cost, performance is encouraging. **Cost improvements came from several fronts: larger scale diluting costs, leveraging idle capacity via the economy network, and ongoing tech and efficiency gains**. **We see no reason this trend cannot persist**, the team is executing well, and there is more to do. **Q:** Can you discuss the Jan 20 change to Brazil's logistics pricing model? Charging different rates by parcel type — what's the macro impact on P&L and margins? **A:** We are differentiating merchant charges. Shipping costs correlate closely with size and weight, **so we switched to a price table based on actual size and weight, aligning charges more tightly with our cost structure**. **It is too early to comment on financial impact**, and we do not provide guidance. We may share more after Q1 based on results. We continuously refine our operating model, and this price-table change is part of that. **Q:** Two on Mercado Pago. First, Brazil deposits grew strongly — are they used to fund credit? If so, what share and what is the impact on NIM? Second, 15–90 day early delinquencies rose despite typical Q4 seasonality; what's the cause and outlook? **A:** On deposits, **we are not using them to fund loans today and are not operating a fractional-reserve model**. We are pleased with deposit growth; since launching 'pots' with higher rates last year, balances rose significantly. Users who adopt 'pots' and keep larger balances show materially higher engagement in Mercado Pago and MercadoLibre. They use credit products, debit cards or apply for cards more often, making deposits a key driver of engagement and NPS. On NPLs, **card delinquency fell to a record-low 4.4% in Q4. The small uptick in NPLs came from consumer and merchant loans**. **NIMAL (net interest margin after losses) matters more than headline NPL, and it improved**, indicating higher profitability. We expanded credit access to more users, including higher-risk segments, but priced for risk and achieved wider spreads QoQ. It was a calculated risk with good outcomes. **Q:** Q4 loan yields rose QoQ while early delinquencies increased. Does that imply 90-day delinquencies may rise over the next one to two quarters? Also, were yield gains driven by Mexico's new risk rather than higher Argentina mix? **A:** We do not provide guidance, but are comfortable with current risk levels. Models are improving, we price risk appropriately and maintain healthy spreads. Expected losses are provisioned upfront, so there should be no surprises. Yields rose similarly across the three countries. We were more cautious in Argentina in Q4 due to election-related macro uncertainty, and higher rates dampened demand. Other markets saw no material change. To add: we only scale credit when the portfolio is healthy. Looking at delinquencies alone is incomplete; Argentina and Mexico margins were far better vs. last quarter. NIMAL, which incorporates losses, provisions and income, improved QoQ, and with confidence in models and collections, our credit book grew 90%. **Q:** On S&M and provisioning. First, marketing ratio rose QoQ; you cited seasonal tactical investments, but Brazil competition seems easing — how should we think about this spend? Second, the acceleration in consumer credit in Brazil and Mexico drove provisioning — was that mainly due to overlap captured via marketplace cross-sell? **A:** **Higher marketing ratio was mainly driven by social-channel expansion**. Our affiliate program is scaling well; in Brazil, affiliates nearly doubled QoQ and rose ~6x YoY in Q4. It is a very promising channel that will drive growth. The ratio sits at the upper end of the 11%–12% historical range, but still within recent investment levels. On credit, split it in two: for high-yield accounts, last year's strong push was the primary driver and had little to do with marketplace activity. For cards, two drivers: we integrated card offers at checkout and expanded installments, making the marketplace a major driver of card usage; and we kept improving model quality, which increased accuracy and confidence to issue more cards. So the primary driver of issuance was model precision, with checkout integration also important. **Q:** Given provisioning and investments to expand the credit book through cards and other loans, can you quantify how cardholders or borrowers spend vs. non-users in AOV or GMV, and their retention? **A:** We do not disclose exact uplift, but the impact is material. We track engagement frequency and NPS, both of which rise meaningfully once users start using our credit products or obtain a card. We also see higher engagement and net spend. Across Brazil, Mexico and Argentina, the share of GMV paid with Mercado Pago products (BNPL, cards, wallet balances, etc.) is rising steadily. These transactions typically have higher approval rates and lower costs, and often generate revenue on the Pago side or at least lower costs. There is strong synergy between the marketplace and Mercado Pago. **Q:** On cards, you issued over 3 mn in Q4 vs. 1.5 mn in Q2. Argentina does not seem the main driver; did Brazil and Mexico step up vs. Q2? As card NPL performance keeps improving, could you lower upfront provisioning, making early margins more attractive? **A:** **From Q2 to Q3, issuance growth was mainly Brazil-driven**, as better models let us find more users who meet our return thresholds. **From Q3 to Q4, two drivers: Argentina, where we began issuing mid-Q3 and accelerated in Q4, adding several hundred thousand cards** with good early results and a focus on low-risk users; and Mexico, where better-than-expected payback drove faster issuance. All three countries are now running at full speed. On profitability, take Brazil's oldest cohort launched in 2021: cohorts with 2+ years are profitable at the NIMAL level, which gives us confidence to expand. Portfolio-wide, we are not profitable yet due to the large influx of new users. We operate three credit lines: consumer loans with high NIMAL (30%–40%+), merchant loans similarly, while cards are not yet positive on average. About half of Brazil's card book is now NIMAL-positive, Mexico is trending well, and it is too early in Argentina. **Q:** On acquiring TPV, how do on- and off-platform volumes affect blended take rate? Is the interchange component in Mexico facing a regulatory cap? **A:** For acquiring, think of three parts: on-platform and off-platform, but on-platform transactions flow into marketplace economics and not recorded in acquiring. In acquiring, we see off-platform online and offline (POS) transactions. **Online carries higher risk and complexity, so rates are higher; POS is very low risk with lower rates**. Online has higher margins, but offline is growing fast and our share is small, so the opportunity is large. On Mexico's interchange cap, it has not passed; the regulator postponed or shelved it. So interchange does not change in the near term. **Q:** On agent commerce, I see the strategic upside and agree it drives online retail penetration and digital ad growth. What risks do you see from independent agents disintermediating customer access and potentially changing monetization, notably ad traffic? How do you plan to respond? **A:** Some things we know, some we do not. We **do not know how people will shop in 10 years or which models will win, but we do know consumers value the best end-to-end experience — not only finding products, but fast delivery, broad selection, price, optimal financing and post-sales support**. We also know current technology can materially improve product discovery. So we are investing heavily in building our own agent and shopping assistant inside MELI. It is too early to judge how other assistants evolve. We understand the risks, but are confident we start from a position of strength: direct consumer relationships, a beloved Latin American brand and historical purchase data to power a great assistant. We focus on what we can control — building the best assistant possible. MELI is well placed to capture this shift. As noted, it will accelerate the move from offline to online retail, which matters in LatAm where online penetration lags the US, UK or Asia by roughly a decade. **** **本文的风险披露与声明：**[**Dolphin Research disclaimer and general disclosure**](https://support.longbridge.global/topics/misc/dolphin-disclaimer) ### Related Stocks - [MELI.US - Mercadolibre](https://longbridge.com/en/quote/MELI.US.md) - [09988.HK - BABA-W](https://longbridge.com/en/quote/09988.HK.md) - [BABA.US - Alibaba](https://longbridge.com/en/quote/BABA.US.md) - [89988.HK - BABA-WR](https://longbridge.com/en/quote/89988.HK.md) - [HBBD.SG - Alibaba HK SDR 5to1](https://longbridge.com/en/quote/HBBD.SG.md) - [GOOG.US - Alphabet - C](https://longbridge.com/en/quote/GOOG.US.md) - [GOOGL.US - Alphabet](https://longbridge.com/en/quote/GOOGL.US.md) - [ROKU.US - Roku](https://longbridge.com/en/quote/ROKU.US.md) - [KBAB.US - KraneShares 2x Long BABA Daily ETF](https://longbridge.com/en/quote/KBAB.US.md) - [BABO.US - YieldMax BABA Option Income Strategy ETF](https://longbridge.com/en/quote/BABO.US.md) --- > **Disclaimer**: This article is for reference only and does not constitute any investment advice.