ShoppingBench: A Real-World Intent-Grounded Shopping Benchmark for LLM-based Agents

TL;DR

ShoppingBench introduces a comprehensive, real-world shopping benchmark with complex user intents, a large-scale simulated environment, and evaluation of language agents' performance, revealing significant challenges and proposing methods for agent improvement.

Contribution

We present ShoppingBench, a scalable, intent-grounded shopping benchmark with a large simulated environment and novel training strategies for improving LLM-based shopping agents.

Findings

State-of-the-art agents achieve under 50% success rate on benchmark tasks.

Our training methods enable smaller agents to perform competitively with GPT-4.1.

ShoppingBench highlights the complexity of real-world shopping tasks for LLMs.

Abstract

Existing benchmarks in e-commerce primarily focus on basic user intents, such as finding or purchasing products. However, real-world users often pursue more complex goals, such as applying vouchers, managing budgets, and finding multi-products seller. To bridge this gap, we propose ShoppingBench, a novel end-to-end shopping benchmark designed to encompass increasingly challenging levels of grounded intent. Specifically, we propose a scalable framework to simulate user instructions based on various intents derived from sampled real-world products. To facilitate consistent and reliable evaluations, we provide a large-scale shopping sandbox that serves as an interactive simulated environment, incorporating over 2.5 million real-world products. Experimental results demonstrate that even state-of-the-art language agents (such as GPT-4.1) achieve absolute success rates under 50% on our benchmark tasks, highlighting the significant challenges posed by our ShoppingBench. In addition, we propose a trajectory distillation strategy and leverage supervised fine-tuning, along with reinforcement learning on synthetic trajectories, to distill the capabilities of a large language agent into a smaller one. As a result, our trained agent achieves competitive performance compared to GPT-4.1.