Pre-release sales forecasting: A model-driven context feature extraction approach

Predictive competitive intelligence with prerelease online search traffic

“…In their studies, they used a variety of PRB sources including forums (e.g., Craig et al., 2015; Liu, 2006), blogs (e.g., Divakaran et al., 2017; Onishi & Manchanda, 2012), Twitter (e.g., Asur & Huberman, 2010; Gelper et al., 2015), and Facebook (Ding et al., 2017; Kim et al., 2017). Another major PRB source is online search traffic available through Google Trends (e.g., Kim, 2021; Kim & Hanssens, 2017; Kulkarni et al., 2012) or Baidu (Tian et al., 2014). Studies that forecast with online information, including PRB, stem from a broad range of disciplines and therefore many lack adherence to well‐established forecasting principles, as highlighted by Schaer et al.…”

“…The most common ways researchers reduce the PRB time dimension are summing its volume over a certain period (e.g., Gelper et al., 2015; Tian et al., 2014; Wang et al., 2010) or capturing its adoption dynamics with diffusion model parameters (Kulkarni et al., 2012) and functional principal components (FPCs; Foutz & Jank, 2010; Hann et al., 2011; Xiong & Bharadwaj, 2014). Although Xiong and Bharadwaj (2014) and Foutz and Jank (2010) directly compared FPC against volume‐based PRB models, they did not include PRB valence; valence and volume are complementary, as they summarize PRB information in different ways.…”

“…Predicting the success of new products prior to launch is a challenging task (Goodwin et al., 2013; Trusov et al., 2013), but recent studies suggest that prerelease buzz (PRB) information can substantially improve new product forecasting (e.g., Kim et al., 2015; Schaer et al., 2019b; Xiong & Bharadwaj, 2014). PRB represents the aggregated anticipated interest of consumers toward a new product (Houston et al., 2018) and has been collected, for example, in the forms of online search traffic information (e.g., Schaer et al., 2019b; Tian et al., 2014), blogs (e.g., Dhar & Chang, 2009; Kim et al., 2015), or microblogs (e.g., Asur & Huberman, 2010; Gelper et al., 2015). It differs from postrelease signals where both pre‐ and postconsumption behavior occurs (Houston et al., 2018).…”