The role of search in online shopping is evolving. Instead of directly entering product names to find desired items, consumers are increasingly relying on artificial intelligence (AI) to interpret their preferences and situations, leading to tailored recommendations. This shift is redefining shopping standards from search-based to recommendation-focused.
According to a shopping behavior analysis report by ResolveAI released on September 26, 70% of shoppers complete their purchases after just one search. Searches consisting of three or more words increased from 25% of total searches in 2024 to 40% in February of this year, while the use of filters and exploration beyond the first page of search results decreased by 13% each. This indicates a shift in the shopping environment where AI interprets consumer intent rather than relying on keyword combinations.
Changes are also evident outside of search bars. A Salesforce survey found that 39% of consumers are using AI for product discovery, with this figure exceeding 50% among Generation Z. A Capgemini study revealed that about three out of five consumers have begun replacing traditional search engines with generative AI. Instead of manually comparing search results, shoppers are now presenting conditions to AI and receiving product recommendations, establishing a new exploration pathway.
This transformation is reflected in platform services as well. Musinsa recently launched a dedicated app within OpenAI's ChatGPT, allowing users to receive product recommendations based on various contexts such as price range and preferred brands, without needing to input specific brand or product names. Previously, Musinsa introduced a service within KakaoTalk that uses AI to recommend styles. A Musinsa representative stated, "We provide refined recommendations through conversational fashion and beauty commerce, and we will continue to drive traffic to the Musinsa store through various service integrations."
Half Club has implemented style-centric AI search, moving beyond keyword-based searches to allow users to find related products using style-focused queries like 'autumn work outfits' and 'trending vacation looks.' The AI analyzes the styles and trends embedded in the search terms and compares them against a database of approximately 7 million products to recommend the most relevant items. The product detail pages also feature an image-to-video technology that transforms static product images into videos, providing a more intuitive view of the product's silhouette and movement, allowing consumers to visualize how items would look when worn. The company plans to expand this feature to include major brand products in the future.
Ably has introduced a generative AI-based virtual fitting service called 'AI Dressing,' enabling users to preview various styles. Consumers can select up to 10 fashion styles they wish to try on and upload a single photo, allowing the AI to superimpose different clothing styles onto their image based on their face and background. This service aims to go beyond mere 'style commerce' to provide engaging content for consumers.
As recommendations become the starting point for shopping, the importance of the product data accumulated by platforms is growing. While AI capabilities are rapidly standardizing, the quality of product information and how well it connects with user preferences will determine the effectiveness of these services. Industry experts predict that AI will expand its role beyond product searches to include comparisons, recommendations, and purchase support.
According to a survey by market research firm Embrain Trend Monitor, 50.2% of respondents expressed interest in trying 'AI shopping agents' but found it difficult to entrust all purchasing decisions to AI. The primary reason cited, by 43% of respondents, was that AI data analysis alone cannot fully grasp individual preferences. As AI recommendation services proliferate, the ability to accurately reflect personal tastes and build consumer trust will be crucial.
Kim Si-wol, a professor in the Department of Consumer Studies at Konkuk University, emphasized, "Since each consumer has different information needs and purchasing criteria, it is essential to build a personalized recommendation system that reflects individual characteristics. The speed at which various information, such as body type and time, place, and situation (TPO), is accumulated and reflected in recommendations will be key." He added, "As recommendation services expand, appropriate accountability systems and refund policies must also be established."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.