Magazine Luiza (MGLUY) WhatsApp AI Assistant Converts 3x Its App

Magazine Luiza S.A. (MGLUY) gave the first public performance read on its Lu WhatsApp AI shopping assistant at its earnings call on 2026-08-07: the conversational channel converts shoppers at three times the rate of the company's own shopping app [1]. How much the channel actually sells is still undisclosed.

Magazine Luiza is one of Brazil's largest retailers. It sells first-party goods through its stores and its own website, runs a marketplace for third-party sellers, and builds logistics, payments, advertising and cloud services around that retail base. Lu is the virtual persona the company has used for years. It now sits inside WhatsApp as a conversational shopping assistant, a channel the company calls "Agente eCommerce" internally: the shopper never has to open the Magalu app, Lu recommends products and helps pick them, identity verification and payment happen inside the same conversation, and post-sale contact stays in that conversation too [1]. Management lists the AI-supported curation experience as one of the priorities of its digital repositioning, and the channel sells first-party e-commerce goods, which places it in the core business [1].

What changed on the 2026-08-07 call

The channel launched at the end of last year and is still ramping. This call moved it to a new place: authentication, checkout and post-sale service now all run inside the conversation, so shoppers no longer have to return to the app. Management said the same flow will be extended to the web in the second half of the year, and ranked it among the handful of items in the company's digital repositioning [1]. Until now the company had discussed Lu mainly as a brand persona and a customer-service entry point. This is the first time it has given a conversion read compared directly against its own app, and the only time it has publicly described what the channel has produced.

How to read the three-times conversion figure

The channel has generated more than 18 million conversations in total, and the conversion multiple is quoted at that scale [1]. The three-times figure has to be read on the basis management gave it: a channel-wide multiple against the shopping app, with no absolute conversion level disclosed for either side and no measurement period stated. It shows that the conversational flow closes a higher share of shoppers than the app does, but it cannot be converted into any amount of money [1].

The gap also cannot be credited entirely to AI. Buying directly inside a WhatsApp conversation already removes two points of drop-off, downloading the app and jumping between pages, and the company does not measure Lu's recommendation step separately. Management separately said that 20% of the customers who bought in the channel came back to buy a second time, while noting that this share is still likely to move significantly, so for now it is only an early reading [1].

What it could move, and what is still missing

If the conversational channel really does close a higher share of shoppers than the app, and if conversation volume keeps growing and the flow reaches the web in the second half, then online e-commerce sales, meaning online GMV and the net revenue carried from it, are the first line that could be pushed up: the same amount of incoming traffic would turn into more orders, buyers would return inside the conversation to buy again, and the company might spend less on reaching those customers a second time. That path is inference at this point. The company has still not disclosed the channel's sales or order volume, has not said what share of its e-commerce sales it represents, and has not attributed any change in online revenue to it. When management explained the recent return to positive online growth, it pointed to a partnership with a third-party platform [1].

What can be confirmed today is a working sales channel with evidence of efficiency behind it: a shopper can select, verify, pay and get post-sale service within a single conversation, and the company's own comparison shows that path closes a higher share of sales than its shopping app. What remains unquantified is how much revenue it brings in. Settling that requires the company to disclose the channel's own sales or its share of online sales, or to give the absolute conversion levels on both sides together with the measurement period behind them.

Application assessment

• Lu WhatsApp Conversational Commerce Assistant | Business position: core business | Deployment stage: limited production | Deployment scope: single business unit | Value type: revenue growth

Sources

[1] Drillr - Magazine Luiza S.A. (MGLUY) - 2026-08-07 - Earnings call

Quote: "We are still with a three times higher conversion than our application and we already have a good number here, 20% of the clients who bought"