Social
E-commerce

A New Paradigm Coupling Social Relationships and Economic Behaviors

Pinduoduo
Beidian
Xiaohongshu
Yunji
Daling Family
Jingxi
Wechat

Introduction

Social commerce typically refers to e-commerce that uses social media to help e-commerce transactions and activities, with classic examples such as Facebook commerce and Instagram commerce. Recently various novel forms of social commerce have become increasingly popular in China, which can be categorized into several types: content-sharing platforms, membership-based platforms, and team purchase platforms. These platforms can achieve great success and expand to a large scale rapidly. For example, Pinduoduo acquired over 200 million users in less than three years and the daily order volume ranks second in mainland China next to Taobao link. Beidian attained nearly 40 million users in less than 1.5 years.

Notably, most recent social commerce platforms are embedded within instant messaging platforms, i.e., WeChat under Chinese context, where the key drivers are mostly real-world relationships, especially close relationships. As such, these platforms penetrate and leverage existing real-world close social relationships and incorporate Chinese guanxi into market transactions, in contrary to traditional social commerce where key opinion leaders (e.g., brand, celebrity) play the leading role. These platforms are typically referred to as social e-commerce (in Chinese,社交电商) in China. Social e-commerce has provided a unique and exciting avenue to investigate the interplay between social relationships and economic behaviors.

Here we present a series of research works to study recent emerging social e-commerce in China from the following two angles:

  • Mechanism understanding
  • Algorithm design

We draw upon an interdisciplinary approach bridging computer science and social science, with researchers from Tsinghua University, Stanford University, Harvard Business School, Kyoto University, etc.

We also released two datasets on social e-commerce to benefit the research community. link

Project I: Mechanism Understanding
Understanding novel characteristics of Social E-commerce

To uncover the characteristics of social e-commerce, we comprehensively investigate social e-commerce from various perspectives. Specifically,

  • We empirically measure network structure and dynamics of social e-commerce platform (Cao et al. ICWSM 2020), where we showed social e-commerce distinguished itself from prior e-commerce in decentralized network structure, invitation cascades, purchasing homophily and user loyalty;
  • We examine purchase motivations and user experiences (Cao et al. Preprint 2020) through in-depth qualitative and study people’s purchase behaviors through large-scale quantitative analysis (Xu et al. CSCW 2019). We find social e-commerce leads to a 3.09∼10.37 times higher purchase conversion rate than the conventional settings, which can be explained by mechanisms of better matching, social enrichment, social proof, and price sensitivity (Xu et al. CSCW 2019). We also showed social e-commerce enables more reachable, cost-reducing, and ubiquitous user shopping experiences, shapes decision-making process and creates novel social interactions (Cao et al. Preprint 2020);
  • To further delineate the ecology of social e-commerce, we illustrate agents’/intermediaries’ roles and uncover agent transformation (Xu et al. ICWSM 2021) and agent role in social e-commerce (Chen et al. CSCW 2020). We find agents on social e-commerce act as local trend detectors and “social grocers” and identify several successful strategies agents take (Chen et al. CSCW 2020). We also identified social conformity, social enrichment, refusal avoidance, and benefit-cost trade-off as mechanisms affecting invitation (Xu et al. ICWSM 2021)

Project II: Algorithm Design
Developing Better Recommender Systems in Social E-Commerce

Based on understanding of social e-commerce mechanisms, we further proposed a series of novel algorithms for building more effective recommender systems on social e-commerce. Specifically, we tackled the following scenarios:

More details about our related publications can be found in link.

Datasets

We open source the following two datasets to benefit the community.

Sharing-Purchase Behavior Dataset

Used in Fine-grained Social Recommendation, Cross-platform Social Recommendation and Word-of-mouth Recommendation

A. Sharing-and-Purchase: Sharer || Receiver || Item
B. Traditional-Purchase: User || Item
C. Social Relation: User || User

Our dataset are available at Github.

Group-purchase Dataset

Used in Group-buying Recommendation

A. Successful Group-purchase Behavior: Launcher || Participant Users || Item
B. Failed Group-purchase Behavior: Launcher || Item
C. Social Relation: User || User

Our dataset are available at Github.

Team

Yong Li

Yong Li
Tsinghua

Fengli Xu

Fengli Xu
Tsinghua

Chen Gao

Chen Gao
Tsinghua

Zhilong Chen

Zhilong Chen
Tsinghua

Jun Zhang

Jun Zhang
Tsinghua

Hancheng Cao

Hancheng Cao
Stanford

Mengjie Cheng

Mengjie Cheng
Harvard Business School

Wang Tao

Tao Wang
Kyoto