Amazon said today that it is using generative AI to help people shop for garments online after using it to improve product ratings. The business is employing massive language models, generative AI, and machine learning to power four AI-powered features that assist consumers in locating clothing that fits, a major issue when shopping online and the primary cause of clothing returns.
Coresight Research found that online clothing returns average 24.4%, eight percentage points higher than the entire internet return rate. Retailers and brands reported more online returns over the last two years. Since home try-ons and shipping are simpler, buyers often purchase an item in many sizes or colors and then return the rest.
Amazon has used AI in four ways to solve this challenge: tailored size suggestions, a “Fit Insights” tool for sellers, AI-powered highlights from customer fit ratings, and rebuilt sizing charts.
Amazon Fashion developed a deep learning system to assist consumers in identifying their best-fitting size across designs using AI for tailored size suggestions.
This approach takes into account brand size systems, product ratings, and consumer fit preferences. The information is merged in real time to advise on the appropriate size for a consumer and adjust as their demands change. It also utilizes AI to find the ideal styles for them. If a family member often purchases for a teen son or daughter, this function might mislead the system as to what size the consumer is. Amazon states that consumers may build different fit preference profiles using the “find your size” tool and that this would not occur if the client was browsing a children’s or adult’s section.
However, “Fit Review Highlights,” another new tool, may assist. This feature expands on the August 2023 AI-generated Customer Review Highlights, which give a brief paragraph overview of customer sentiment and product characteristics with clickable links for important product aspects.
Amazon uses user reviews to extract fit information, including size accuracy, garment fit on particular body parts, and fabric stretch, for Fit Review Highlights. Amazon says large language models collect information from customer evaluations, and AI summarizes the results into an easy-to-read highlight specific to the user. This might save consumers time by eliminating the need to read hundreds of reviews to determine fit.
The store is also using AI to enhance site size charts. Amazon Fashion extracts and cleans product size charts from numerous sources and standardizes them using large language models. Remove redundant information and auto-correct missing or erroneous measurements for more accurate charts and fits.
AI-powered insights will benefit sellers too. Amazon Fashion’s Fit Insights Tool helps merchants understand consumers’ fit demands and convey sizing, such as “true to size,” or if an item runs smaller or bigger.
Amazon says they can utilize this data to direct future production. In this scenario, big language models capture and aggregate client comments on fit, style, fabric, returns, size chart assessments, and reviews. The brand’s size charts are checked for inaccuracies using machine learning.
Amazon has made many AI improvements to enhance the shopping experience on its site in recent months. In addition to user reviews, Amazon introduced generative AI capabilities to help merchants create product descriptions and improve product photos. The retailer anticipated 40% higher click-through rates with the latter. Amazon added AI to Alexa and Fire TV in addition to Web Services.
