The Role of Color Complexity and Socialness of Listings in Crowdfunding Success

Listing images provide important initial unstructured information on candidate projects to potential backers on crowdfunding platforms. Color psychology theory suggests that complex images containing a diverse palette with very different colors are likely to be less attractive to individuals. However, listings that convey social cues are likely to be more attractive through the generation of a social connection with the user. This paper tests the predictive value of a model combining variables from both theoretical perspectives to predict crowdfunding success. The study uses a data set of 176,614 projects from Kickstarter. Listing images were analyzed using various AI and machine learning techniques. Several text features were also extracted from the listing text. The predictive model used logistic regression, while controlling for a variety of additional factors, and found support for the influence of socialness and color complexity on crowdfunding success. The model has considerable predictive value and explained 56.4% of variance, with an area under the curve of 94%, and an accuracy of 85%. Implications for future research and practice are explored.