Customer segmentation for stigmatized products using point-of-sale big-data and Gaussian-Mixture-Model: an application in female-hygiene market

Understanding customers of female hygiene products remains a challenge despite the market’s size and social importance, due to taboos and limited reliable data. This study proposes an alternative approach using retail scanner data and latent class segmentation based on Gaussian Mixture Models (GMM). Analyzing 260 behavioral metrics from 2497 customers, 582 stores, and over 276,000 transactions, we identified four distinct consumer segments with 28 significantly different behavioral characteristics. To profile these segments, we employed machine-learning-based classification models, revealing clear demographic patterns and the influence of female life stages on purchasing behavior. Additionally, we calculated the Customer Lifetime Value (CLV) for each segment, uncovering variations in segment profitability. Our approach offers a scalable, behavior-driven method to segment customers without relying on intrusive personal data. This enables marketers to design hyper-personalized experiences and more inclusive targeting strategies for both active and latent consumers of feminine hygiene products. The findings also highlight untapped consumer opportunities beyond the traditionally targeted coming-of-age segment, encouraging marketers to adopt nuanced, data-informed engagement strategies that promote both profitability and social inclusivity.