Potential buyer identification and purchase likelihood quantification by mining user-generated content on social media

Abstract Understanding the purchase likelihood of potential buyers is an important prerequisite for marketers to carry out targeted marketing. Massive authentic and personalized user-generated content (UGC) generated on social media, reflecting the content creator’s purchase intent, provides a new possibility for decision-makers to accomplish this task yet remain mostly untapped by many firms. As such, the current research develops a two-stage approach where potential buyers are first identified based on the premise of classifying user's posts into before buying and after buying, and their purchase likelihood is quantified by a novel Weighted Recency, Focus, and Sentiment (WRFS) model. Data from the Honda Civic community in the Bitauto automotive forum are employed to verify the proposed method. 2492 from 10,229 users in the Honda Civic community were identified as potential buyers, and their purchase likelihood is obtained by the WRFS model. The actual purchases of these potential buyers are then observed and verified. The results highlight that the higher the purchase likelihood, the higher the proportion of users who purchase, which illustrates the accuracy of the proposed method.