Consumer segmentation with large language models

Abstract

Consumer segmentation is vital for companies to customize their offerings effectively. Our study explores the application of Large Language Models (LLMs) in marketing research for consumer segmentation. We developed a workflow leveraging LLMs to perform clustering analysis based on consumer survey data, with a focus on text-based multiple-choice and open-ended questions. Firstly, we employed a LLMs model to embed text for clustering, demonstrating that LLMs enhance clustering accuracy over traditional models. Secondly, we created persona chatbots using LLMs, which achieved over 89% accuracy in simulating consumer preferences. Our findings underscore the potential of our LLMs framework in marketing research.