Automated Video Analytics in Marketing Research: A Systematic Literature Review and a Novel Multimodal Large Language Model Method.
Type
forthcoming
Date Issued
2025-05-21
Author(s)
Abstract
The exponential growthof online video content presents significant opportunities for marketing researchersto analyze online consumer behavior. However, effectively usingthis data requiresadvanced methods for managing the complexity and sheer volume of video data. This researchpresents a systematic literature review of video analytics in marketing research, outlining the types of information extracted automatically from video data and examining the specific methodologies employed by researchers.Furthermore, we evaluate a multimodal large language model, specifically ChatGPT-4o, for complex, zero-shot imagecoding tasks in marketing research. Building on theseimage coding capabilities, we introduce a novel multimodal large language model-based pipeline for zero-shotvideo analysis. This innovative method allows researchers to efficiently extract meaningful informationfrom video data without prior training or labeled datasets. Our findings highlight how multimodal large language models can advance video analytics into a scalable andcost-effective tool for marketing research.
Language
English
HSG Classification
contribution to scientific community
Refereed
No
Official URL
Subject(s)
Division(s)