Analyzing and Predicting Consumer Response to Short Videos in E-Commerce

Abstract

This study analyzes the drivers of and predicts the outcome of consumer response to e-commerce short videos (ESVs) in terms of viewing duration. We first construct a large-scale ESV dataset that contains 23,001 ESVs across 40 product categories. The dataset consists of the consumer response label in terms of average viewing durations and human-annotated ESV content attributes. Relying on the constructed dataset, we employ econometric modeling and deep learning methods to comprehensively understand and predict consumer response to ESVs. First, by employing the mixed-effects model, we find that ESV content attributes for product description, product demonstration, pleasure, and aesthetics are key determinants of ESV viewing duration. Subsample analyses further show the heterogeneous effects of different content attributes on ESV viewing response across product categories. Second, we design a content-based multimodal-multitask framework to predict consumers’ viewing response to ESVs. We propose an information distillation module to extract the shared, special, and conflicted information from ESV multimodal features. Additionally, we employ a hierarchical multitask classification module to capture the feature-level and label-level dependencies. By conducting extensive experiments, we demonstrate that the prediction performance of our proposed framework is superior to that of other state-of-the-art models. Taken together, our paper provides novel theoretical and methodological contributions to the Information Systems and relevant literatures.