Amazon Fashion uses AI to improve online clothing shopping

The company developed a learning-based algorithm that aims to support customers in finding their best-fitting size in all styles, as it considers the sizing relationships between brands and their size systems, a product’s reviews, and customer’s fit preferences. The algorithm anonymously groups customers with similar size and fit preferences, and products with a comparable fit. The solution then learns from millions of product details, including style, size chart, and customer reviews, as well as anonymised customer purchases, while taking into consideration sizes previously acquired and kept by similar customers for the same product. Afterward, the feature recommends in real-time the suitable size for a customer. Amazon’s algorithm also considers the fluctuations in sizes across time, with the company developing it to continuously learn and self-adapt to these modifications. Additionally, the company leverages AI to support customers in discovering alternative styles based on their preferences. The AI-enabled recommendations isolate product data, including style, colour, price, size, rate of return, and customer reviews from the product catalogue to endorse other styles as the user shops.

Furthermore, Amazon introduced an AI-generated Fit Review Highlights feature that develops a review highlight for each customer considering their suggested size using mutual themes across evaluations. The newly added feature informs the customer whether to size up or down in a style based on reviews from other users who bought the item in the same size. Amazon leverages large language models to extract data from customer reviews, such as size accuracy, fit on specific body areas, and fabric stretch. Details are then summarised using AI in a user-friendly review highlight that guides customers to the relevant information. Size charts were also improved by using large language models that automatically extract product sizes, remove duplicate information, and auto-correct missing or incorrect measurements, making them more accurate and consistent. The company is currently experimenting with additional methods of providing the most relevant size and measurements for each user, including grouping measurements for their respective size.

In addition to providing customers with enhanced services, Amazon supports brands and selling partners with the new features. Leveraging a large language model to extract and aggregate customer feedback on fit, style, and fabric, the company’s Fit Insights Tool allows sellers to receive contextualised information on why a product was returned. The feature uses machine learning to identify defects in size charts. Brands can use this data to understand customer fit issues, enhance their size communication, and incorporate feedback into future designs and manufacturing.

Source: Link