“Leveraging Unstructured Big Data to Understand Consumption Journey” by Dr. Wenbo WANG
Dr. Wenbo WANG
Department of Marketing
The Hong Kong University of Science and Technology
Consumption of experience goods such as movies, theme parks, sports is often a course of complete journey from beginning to end, rather than a mere point of purchase outcome. For example, a movie viewer’s experience during two-hour watching can vary significantly across the storyline of a movie. Real time user generated content (UGC) data on social media, such as pictures, tweets, short videos, emoji makes it possible for firms to decode the consumption journey of consumers in a low-cost scalable way. This research proposes a novel concept, moment-to-moment synchronicity (MTMS), to measure real time experience along the consumption journey. Specifically, MTMS refers to the synchronicity between the product offerings and consumer moment-tomoment reactions. In the movie category, for example, a high level of MTMS means that viewers closely follow the “rhythm” of the movie along the timeline.
We validate this concept using online movie watching as a demonstration case. We investigate consumer’s moment-to-moment movie viewing experience by analyzing real-time on-screen UGC. On-screen UGC are live chats, typed in by online viewers while watching a video, then displayed on-screen simultaneously with the video stream, and can be seen by future viewers watching the same video. We collect and merge unique data of on-screen UGC, movie stream data including video, audio and subtitle, as well as quality evaluation of movies.
We first use a series of state-of-the-art computer science algorithms to convert video, audio and subtitle of a movie into numerical data. Specifically, along the timeline of each movie, we quantify second-by-second movie contents including camera shot transition, camera motion, sound energy, pitch level, number of dialogs in subtitles, and number of exclamation tone among the dialogs. MTMS of a movie is then measured by the R-squared in the regression of UGC volume onto movie content variables. We find that MTMS significantly predicts quality evaluation of a movie. This result is consistent to the intuition that consumers like a movie when their real-time viewing experiences highly co-move with the rhythm of the movie.
This research has the following contributions. First and theoretically, the proposed concept of MTMS links consumer real time responses to product offerings. Our findings on the relationship between MTMS and quality evaluation lends theoretical support to recent neuroscience research on neural responses on video stimuli. Second, we add to the UGC literature by introducing a new form of UGC that are highly disaggregated at time level to tape moment-to-moment consumer reactions. On the down-stream effects of UGC, we show that the distributional feature of UGC volume during consumption predicts the overall quality evaluation of a product. Third, our research adopts a series of state-of-the-art computer science tools to extract quantitative data from multi-media data streams. We introduce several scalable multi-media analysis techniques to marketing research. Finally and substantively, the proposed MTMS provides companies with new insights on movie production.