[DigitalToday reporter Dae-geon Seok (석대건)] The axis of competition in online shopping mall advertising is shifting from creation to operations. While generative artificial intelligence (AI) has lowered barriers to making creative materials, performance gaps are widening at the operations stage, where results are analysed and budgets are reallocated after execution. Cafe24 is responding to this trend with its next-generation marketing solution NextGen, which links the entire advertising operations process with data and AI.

Advertising for online shopping malls does not end once it is launched. Operators must keep checking which products sell, which ads lead to actual orders and where to add more budget. They must stop underperforming ads, replace creative materials and frequently adjust strategies by comparing performance across multiple media channels.

As a shopping mall grows, these tasks become more complex. As the number of products and ad channels to manage increases, so do the decisions to make. Ads run around the clock, but an operator cannot watch them all day. Budgets are spent in the meantime, and ads whose performance has declined continue to be shown. That is why success hinges less on a single good ad than on how quickly operations can be repeated.

◆Gartner: marketing automation share to reach 36 percent in 2028

AI's role is also shifting accordingly. If early generative AI focused on creating ad copy and images, it is now expanding toward analysing performance, proposing operating strategies and automating repetitive tasks.

A Gartner outlook survey of marketing leaders released in 2026 forecasts the share of marketing tasks automated by AI will expand to 36 percent in 2028 from 16 percent in 2026. That means AI is evolving beyond creation tools toward supporting marketing operations overall.

NextGen, introduced by Cafe24's marketing centre, also targets this point. The company says the core is not executing ads on behalf of users but connecting the entire operations process with data and AI. It analyses performance in real time, compares data across multiple ad platforms and supports budget adjustments and strategy building. It automates repetitive tasks, and operators can analyse performance or request strategies using only natural language.

Even so, AI support for operations does not automatically produce results. For AI to properly analyse performance and propose strategies, it must also make judgments based on sufficient data.

A Bain & Company report released in 2025 found that companies that actively use AI to deliver results had higher levels of data infrastructure and system integration. Some 66 percent of surveyed companies cited data infrastructure and system integration as a key challenge in using AI. It said AI performance depends less on the model itself than on how effectively data is connected and used.

The same context applies to advertising. Only by connecting performance data scattered across multiple platforms into a single flow can operators make comprehensive decisions on which products to focus ads on, which platforms to allocate more budget to and which creative materials to scale up. This is the background to NextGen choosing a structure that integrates and analyses data by platform.

◆Moreout ROAS 540 to 759 percent, Kikiko ad spend down 12 percent

Cafe24 says NextGen's effects are confirmed in actual operating cases. Lifestyle yoga-wear brand Moreout continuously tested AI-based ad creative and analysed product data together with ad performance to select products to focus advertising on. As a result, its Meta return on ad spend (ROAS) rose to 759 percent from 540 percent, and Meta conversion sales also increased 33 percent.

Teen girls' apparel brand Kikiko used NextGen's real-time performance analysis and automatic optimisation to cut ad spending 12 percent in the off-season while lifting ROAS to 855 percent from 579 percent. It was a case of improving efficiency without increasing ad spending.

Cafe24 says what the two cases share is that AI did not execute ads on their behalf, but analysed data and continuously proposed strategies during operations. It can be seen to mean that what lifted performance was not more ad spending but repeated, data-based operations.

Reproducibility remains an obstacle, but the overall trend is changing. Whether the performance of a specific brand appears the same for sellers in different industries and of different sizes remains a task. Even so, as the next stage of advertising competition shifts from making more ads to how quickly data is read and turned into execution, the standard for shopping mall operators' competitiveness is also moving.