Tailor: Size Recommendations for High-End Fashion Marketplaces

TL;DR

The paper tackles size recommendations in luxury fashion marketplaces, where wrong sizes lead to costly returns. It introduces the Sequence Size Predictor (SSP) family, including SSP-LSTM and SSP-Attention, which encode user event histories into fixed-length representations to predict a size distribution $p(y|x)$ for a given (customer, product) pair. By incorporating implicit Add2Bag interactions and explicit ReturnReason feedback, the approach achieves substantial improvements over baselines such as SFNet (up to 45.7% in accuracy) and boosts user coverage (about 24.5%) while maintaining real-time latency. The work demonstrates the practical viability of sequence-based, signal-rich size recommendations in high-end marketplaces and outlines directions for richer product and customer signals to further enhance performance.

Abstract

In the ever-changing and dynamic realm of high-end fashion marketplaces, providing accurate and personalized size recommendations has become a critical aspect. Meeting customer expectations in this regard is not only crucial for ensuring their satisfaction but also plays a pivotal role in driving customer retention, which is a key metric for the success of any fashion retailer. We propose a novel sequence classification approach to address this problem, integrating implicit (Add2Bag) and explicit (ReturnReason) user signals. Our approach comprises two distinct models: one employs LSTMs to encode the user signals, while the other leverages an Attention mechanism. Our best model outperforms SFNet, improving accuracy by 45.7%. By using Add2Bag interactions we increase the user coverage by 24.5% when compared with only using Orders. Moreover, we evaluate the models' usability in real-time recommendation scenarios by conducting experiments to measure their latency performance.

Figures (3)

• Figure 1: SSP-LSTM architecture
• Figure 2: SSP-Attention architecture
• Figure 3: Model Latency Benchmark