Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Su, Xuanbo, Hu, Wenhao, Su, Haibo, Chen, Yunzhang, Zhan, Le, Yang, Yanqi, Huang, Leo
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2604.07054
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914460982050816
author Su, Xuanbo
Hu, Wenhao
Su, Haibo
Chen, Yunzhang
Zhan, Le
Yang, Yanqi
Huang, Leo
author_facet Su, Xuanbo
Hu, Wenhao
Su, Haibo
Chen, Yunzhang
Zhan, Le
Yang, Yanqi
Huang, Leo
contents Sales dialogues require multi-turn, goal-directed persuasion under asymmetric incentives, which makes them a challenging setting for large language models (LLMs). Yet existing dialogue benchmarks rarely measure deal progression and outcomes. We introduce SalesLLM benchmark, a bilingual (ZH/EN) benchmark derived from realistic applications covering Financial Services and Consumer Goods, built from 30,074 scripted configurations and 1,805 curated multi-turn scenarios with controllable difficulty and personas. We propose a fully automatic evaluation pipeline that combines (i) an LLM-based rater for sales-process progress,and (ii) fine-tuned BERT classifiers for end-of-dialogue buying intent. To improve simulation fidelity, we train a user model, CustomerLM, with SFT and DPO on 8,000+ crowdworker-involved sales conversations, reducing role inversion from 17.44% (GPT-4o) to 8.8%. SalesLLM benchmark scores correlate strongly with expert human ratings (Pearson r=0.98). Experiments across 15 mainstream LLMs reveal substantial variability: top-performance LLMs are competitive with human-level performance while the less capable ones are worse than human. SalesLLM benchmark serves as a scalable benchmark for developing and evaluating outcome-oriented sales agents.
format Preprint
id arxiv_https___arxiv_org_abs_2604_07054
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Sell More, Play Less: Benchmarking LLM Realistic Selling Skill
Su, Xuanbo
Hu, Wenhao
Su, Haibo
Chen, Yunzhang
Zhan, Le
Yang, Yanqi
Huang, Leo
Computation and Language
Sales dialogues require multi-turn, goal-directed persuasion under asymmetric incentives, which makes them a challenging setting for large language models (LLMs). Yet existing dialogue benchmarks rarely measure deal progression and outcomes. We introduce SalesLLM benchmark, a bilingual (ZH/EN) benchmark derived from realistic applications covering Financial Services and Consumer Goods, built from 30,074 scripted configurations and 1,805 curated multi-turn scenarios with controllable difficulty and personas. We propose a fully automatic evaluation pipeline that combines (i) an LLM-based rater for sales-process progress,and (ii) fine-tuned BERT classifiers for end-of-dialogue buying intent. To improve simulation fidelity, we train a user model, CustomerLM, with SFT and DPO on 8,000+ crowdworker-involved sales conversations, reducing role inversion from 17.44% (GPT-4o) to 8.8%. SalesLLM benchmark scores correlate strongly with expert human ratings (Pearson r=0.98). Experiments across 15 mainstream LLMs reveal substantial variability: top-performance LLMs are competitive with human-level performance while the less capable ones are worse than human. SalesLLM benchmark serves as a scalable benchmark for developing and evaluating outcome-oriented sales agents.
title Sell More, Play Less: Benchmarking LLM Realistic Selling Skill
topic Computation and Language
url https://arxiv.org/abs/2604.07054