A small series of serendipity Balancing serendipity in the algorithmic recommendation design of video-on-demand layouts based on consumer characteristics

Recommendation systems are a common tool to guide consumers through information and product overload in online environments. The evolving market of video-on-demand (VOD) embraced the prospect of recommendation systems and is continuously searching to enhance its performance. The implementation of serendipity in recommendation systems is increasingly linked as a solution for the issue of filter bubbles and to assess and evaluate user satisfaction within VOD platforms. However, the concept of serendipity introduces complexity in analyzing its implementation, due to its subjective essence and absence of an academic definition and measurement. This thesis provides a literary foundation on recommendation systems, consideration of particular consumer characteristics in user profiles, and a definition and measurement of serendipity. Therefore, this thesis’s objective is to establish the role and level of serendipity in VOD environments for consumers that present particular characteristics by following the research question: To what extent do users perceive and are affected by serendipity in VOD layouts?. Similar to correlated literature, a quantitative approach is employed with data collecting through the distribution of a survey that includes a quasi-experiment. The data is gathered amongst VOD consumers that possess a user profile without the interference of others. Statistical analysis is performed with the help of SPSS and found that two serendipity items, instead of three, are applicable in the research design. The serendipity elements of novelty and unexpectedness are combined, while relevance is separately considered. Concluded from the insignificant paths between consumer characteristics and both serendipity elements, the findings of this thesis indicate the inability of consumers to perceive serendipity in their personalized VOD environment and a 100% serendipity stimulus. The main results indicate that the serendipity component of relevance records the most substantial mediating effect on the performance of the recommendation systems, measured by means of user satisfaction. To known knowledge, this thesis is the second attempt that considers the need for serendipity for specific consumer characteristics in VOD environments. By including the presented consumer characteristics in user profiles, the coping ability and need for serendipity are reflected in the algorithmic design and, therefore, the personalized VOD interface. The implementation of serendipity based on consumer characteristics helps consumers to broaden their preferences and VOD companies to increasingly set foot in the evolving market.