The role and presence of online videos has recently grown significantly. A high number of renowned consulting houses and companies report the current and future growth of their importance concerning digital marketing activities. This new trend will create a new wave of video content produced by users and companies in order to capture the attention of their audiences. Both scientific and industry-specific sources inform that there are many various features of videos that contribute to their positive perception by those audiences. In this master thesis, we look at videos on Kickstarter - a crowdsourcing platform where videos are used to tell a story about products or ideas, thus promote them to a broader audience. The video must appeal to the users who, among other factors, base their decision to support a project on the platform financially, and at the same increase its chance to become successful. We focus our work on 600 project videos to examine their intrinsic properties that make them more or less impactful in combination with specific project variables such as funding goal. The objective of this study is to find out what Kickstarter project video properties and qualities contribute to its successful or unsuccessful outcome. Firstly, we derive from the data science field to collect and analyze the video data with the use of Python scripts; we receive data about the videos: their length, scenes number and duration, and the facial emotions present in them. Secondly, we employ machine learning techniques to engage in an experiment with an aim to identify variables and correlations that can contribute to the funding success or failure of a project. Thirdly, we analyze and visualize the data that allows us to draw conclusions and propose recommendations for further research or production of higher quality user-engaging video content; all the observations and conclusions pertain only Kickstarter technology project videos. We derive from different theoretical angles and academic findings; the methodology pillars are research methods such as CRISP-DM model and Layers of Research Design. This paper also draws from literature connected with video analysis and crowdsourcing to better justify its timeliness and close association with the study program. The results suggest that facial emotions in videos play a less significant role than the video properties: number of scenes or video length. At this stage the biggest predictor for a campaigns outcome by far was the project updates, as they send the strongest signals of engagement between creators and supporters on Kickstarter.
|Educations||MSc in Business Administration and Information Systems, (Graduate Programme) Final Thesis|
|Number of pages||68|