Virtual shoppers instead of panels
According to The Korea Herald, South Korean retailers are testing products on AI-generated buyers before bringing them to real customers. These 'synthetic consumers,' built on large language models and consumer data, are designed to imitate real purchasing behavior and allow companies to test products, marketing campaigns, and store layouts without the time and cost of traditional surveys. The novelty is not only technological: it changes how businesses collect feedback before investing in production or advertising.
The starting point is classic demographic personas, such as 'woman in her thirties' or 'man in his sixties.' The article explains that synthetic consumers go beyond this, incorporating purchasing behaviors and preferences to allow companies to interact with a virtual customer and evaluate an idea before launch. The information is attributed to the South Korean newspaper.
From the online store to the physical store
In July, Lotte Members began developing an 'AI Twin Consumer' with local companies Vibe Co. and Zero to One Partners, combining purchase data from about 45 million users accumulated in the L.Point loyalty program with social media data. The stated goal is to understand not only what customers buy but also the reasons for and context of consumption.
On the e-commerce front, Coupang and Naver are testing synthetic consumers for market research and user experience testing: the technology simulates navigation in an online store, reactions to product information, and the points where the customer encounters friction. BGF Retail, which operates the CU convenience stores, is extending the concept to physical retail outlets: after using synthetic consumers to evaluate new product concepts, it is working on virtual simulations of the store to test merchandise displays, customer flows, and assortments before modifying real stores.
The adoption is not limited to retail distribution. The article cites food companies CJ Group, Daesang, Maeil Dairies, Pulmuone, and Sangmidang Holdings among participants in pilot projects with Intellicia, a synthetic consumer technology provider, which according to the company has completed 120 projects with 50 businesses.
The promises and limits of simulation
Intellicia claims that its synthetic consumers are generated from real interviews, surveys, commercial reviews, and social data, rather than from simple assigned demographic characteristics. The company reports a 91.2% correlation with a survey of 500 real consumers in its own benchmark tests: this is a company claim, reported by the source, not an independently verified figure.
A retail company official quoted in the article observes that in a market where trends emerge and disappear within weeks, synthetic consumers offer insights without having to recruit real panels each time. The same source predicts that in the future companies will be able to make faster decisions based on large groups of virtual customers. Still to be verified are the reliability of the models across different product categories, the handling of personal data used to build the digital twins, and the extent to which simulations will truly replace research on real consumers.
