Meta’s Muse Reaches 10 Million Users in Six Weeks. Can Naver and Kakao Do the Same?

Expectations for AI shopping agents are growing after Muse, Meta’s artificial intelligence (AI) shopping assistant, surpassed 10 million users in the United States and Canada just six weeks after its launch. Naver and Kakao, which are also pursuing shopping agents, are working to strengthen their services. Naver, which already operates an AI shopping agent, is enhancing the service, while Kakao plans to begin a closed beta test of “Kanana Space” in December. However, a growing user base for shopping agents from Meta, Naver and Kakao does not automatically translate into revenue. To expand the range of products consumers can compare, the agents need to connect with external shopping platforms. The companies must give those platforms a reason to participate and decide how to share revenue generated by transactions. Connecting an AI shopping assistant to external shopping platforms can expand consumer choice, but it may reduce a platform’s existing transaction revenue. / Image generated by ChatGPT Platforms that operate their own shopping businesses must also consider how recommending products sold elsewhere could affect existing transactions. If they recommend only products from sellers on their own platforms, comparisons with external shopping platforms are limited. If they also recommend external products, consumers have more choices, but purchases that would otherwise have taken place on the platform could move to competitors. Expanding the range of recommendations may benefit consumers without necessarily benefiting platform revenue. Whether external shopping platforms participate is another variable. Having AI find and compare products across multiple platforms is different from connecting users to purchasing and payment functions. The parties must separately coordinate whether the AI can access product information, whether purchasing functions can be integrated and who bears the costs when a transaction occurs. In particular, shopping platforms with their own customers and seller networks need a reason to provide product and transaction information to an external AI service and pay fees on top of that. Meta, too, is at a stage where it must turn Muse’s growing adoption into revenue. At its earnings announcement on Oct. 21, local time, Meta disclosed usage figures for Muse and identified service expansion, monetization and improved computing efficiency as tasks ahead. Naver and Kakao are expanding their AI agents from different starting points. Naver is focusing on using data from its own services, including shopping and place-related data. Kakao is focusing on expanding partnerships with external businesses. As a result, the two companies could differ in how they secure products and services to recommend, as well as in how they share revenue. Naver operates an AI shopping agent based on Naver Plus Store. Kakao is preparing Kanana Space, which performs tasks in KakaoTalk in response to user requests. Kakao plans to expand its functions to include ordering, gifting, reservations and money transfers. To earn fees by connecting users to external transactions, an AI agent must demonstrate that it contributes to additional sales or lower costs. From a shopping platform’s perspective, there is little incentive to partner if the AI merely changes the route of a purchase that would have occurred anyway while adding a fee. On the other hand, there is room to share revenue if the AI attracts new customers, increases purchase conversion rates or reduces marketing costs. “To secure transaction fees through AI agents, companies must provide tangible benefits to external businesses,” said Kim Dae-jong, a professor in the Department of Business Administration at Sejong University. “Companies like Amazon that have their own platforms and customer data have little incentive to pay fees to Meta.” He added, “To collect fees through an AI agent, it must use AI to increase consumers’ purchase conversion rates and reduce companies’ marketing costs.” Platforms that operate their own shopping businesses must also consider the gains and losses associated with recommending external products. If a purchase that would have taken place on their platform moves to another shopping platform, sellers on the platform could lose sales opportunities, and the platform could lose the revenue it would have earned from that transaction. However, if external recommendations increase service usage and generate new transactions or partnership revenue, those gains could offset the losses. In other words, revenue from new transactions and partnerships must exceed the losses caused by existing transactions moving elsewhere. Limiting recommendations to products within a platform also has drawbacks. Even if the AI selects a suitable product from within the platform, it does not establish whether the price and purchasing conditions are better than those offered by external shopping platforms. Consumers may need to use another service or search on their own to find better terms. “Even if AI makes product searches more convenient, consumers may struggle to obtain enough information to compare prices if the product information available on each platform is limited,” said Lee Jang-seok, a professor in the Department of Media and Communication at Gachon University. “Even the lowest price recommended by AI may be limited to products sold on that particular platform.” He added, “Just as users must subscribe separately to different streaming services to watch each platform’s original content, consumers may have to use multiple AI agents or conduct additional searches themselves if product information is divided among platforms.” jubar@chosunbiz.com