Social media analysis offers businesses a new way to forecast feature demand

Social media analysis offers businesses a new way to forecast feature demand Sadie Harley Scientific Editor Andrew Zinin Chief Editor Social media has become an important source of information for consumers making purchasing decisions, with prior research finding that more than 90% of consumers consult social media before making a purchase. For businesses, that makes social media a potential source of information about what consumers are looking for in products. New research from the University of Florida Warrington College of Business finds that social media posts that evoke consumers' purchase intentions can provide businesses with signals of demand for specific product features and are associated with higher product sales. The study is published in the journal Production and Operations Management. "To make decisions, firms today require interpretable metrics that link online content to operational actions," said Liangfei Qiu, Ph.D., study co-author and PricewaterhouseCoopers ISOM Professor. "Our research findings fill an important gap between predictive analytics and managerial decision-making, enhancing both the usability and impact of data-driven tools in practice." Through the development of an AI deep learning framework, the researchers analyzed both the text and images in more than 3 million Instagram posts and sales data from a kitchenware manufacturer to identify content that could evoke consumers' purchase intentions. They then used those posts to create a measure called purchase-evoking frequency (PEF), which captures how often posts associated with purchase intentions mention a particular product feature. PEF, they found, had a significant positive effect on product sales. Using color as an example product feature, the researchers examined products offered in multiple colors. They found that when purchase-evoking social media posts more frequently mentioned a particular color, sales of products associated with that color were higher. The relationship between PEF and sales varied depending on the type of product and where it was sold. For example, the effect of color PEF was stronger for search goods than experience goods. Search goods are products whose features can be evaluated before a purchase, while experience goods are products for which customers place greater value on functionality and intrinsic quality. The researchers also found that the effect of color PEF was stronger for online sales than offline sales. Based on their findings, the researchers note that businesses can use these feature-level signals to support product design decisions, inventory planning and supply chain coordination. "Our findings suggest that businesses can use social media to identify specific product features that are gaining traction among customers, rather than relying only on broad measures of social media engagement such as likes or shares," Qiu said. More information Zhechao Yang et al, Identifying purchase-evoking social media posts: A theory-driven deep multimodal learning framework, Production and Operations Management (2026). DOI: 10.1177/10591478261474393 Provided by University of Florida