Success factors of crowdfunding projects: a predictive model based on binary logistic regression

Problem. The absence of reliable tools for predicting the success of crowdfunding projects at the early stages of campaigns limits the ability of platforms to select promising initiatives and reduces investment efficiency. Existing research is fragmentary and does not account for the specifics of different project categories or the characteristics of the Russian market. Aim. The aim of this work is to construct a predictive model based on binary logistic regression that allows assessing the probability of a project’s success on a crowdfunding platform. Methods The study investigated the influence of the following factors on the success of crowdfunding projects: target amount and fundraising duration, number of sponsors and average contribution size, presence of a video file in the description, number of news updates and comments, author’s experience, use of social networks; number of project subscriptions; number of days required to raise 25 % of the target amount; and the number of “quick” investments. Additionally, the analysis was conducted considering the project category. The study employed binary logistic regression as a predictive analysis method. Results. The study showed that for predicting the success of most projects, a single factor (the presence of more than one “quick” investment) is sufficient, and the accuracy of such prediction is very high. For example, for projects in the “Creative Products” category, the success prediction accuracy was 99.42 %, and the failure prediction accuracy was 98.98 %. Conclusions. The binary logistic regression model built on data from the Russian platform Planeta.ru allows for high-accuracy prediction of crowdfunding project success. Key predictors are raising 25 % of the target amount within the first week and the presence of “quick” investments. For categories with low success rates (“Business”, “Innovations”), accuracy increases when combining factors. The results can be used by platforms for scoring and project support.