AWS Launches Virtual Try-On Feature to Combat $890B Retail Returns Problem

AWS Launches Virtual Try-On Feature to Combat $890B Retail Returns Problem

Amazon Web Services introduced a virtual try-on capability within Amazon Nova Canvas designed to reduce online fashion returns, which account for one in four clothing purchases. The feature uses two-dimensional image inputs—a source image of a person and a reference product image—to generate realistic try-on previews.

The move targets a structural pain point in e-commerce: one-quarter of online clothing purchases are returned, with poor fit, wrong size, and style mismatch cited as top reasons. Each return generates 30% more carbon emissions than the initial delivery and delays inventory reprocessing. AWS noted that high-value customers often return items most frequently, forcing retailers to maintain costly return policies.

Early virtual try-on solutions struggled with accuracy and scalability, particularly in preserving garment details like draping, patterns, and logos. Amazon Nova Canvas addresses these technical constraints through its generative capability, which AWS says handles these details more effectively.

AWS is publishing sample code and implementation guidance in a two-part series, with part two covering real-world applications and business benefits. The feature is available now within the Nova Canvas service on AWS.

For CFOs evaluating customer acquisition costs against return processing expenses, this represents a potential lever on both margin and customer lifetime value—if accuracy holds in production environments.