Referral Program Optimization Using Machine Learning Based on Promoty’s Case
Name
Marelle Ellen
Abstract
Referral programs are one of the best options for startups to achieve fast growth. Despite that, many Estonian startups have not managed to run successful referral programs. For that reason, a methodology was proposed for Estonian startups for applying supervised and unsupervised machine learning to analyze and optimize their referral program. The methodology was implemented on a startup called Promoty. For every Instagram user not connected to Promoty, the model estimates the probability of becoming so-called „useful user“ who invites at least one new member to Promoty’s platform. With 60% precision, the model is able to identify 23,6% of all the useful users. Therefore, additional features that should be added to improve the model’s prediction power were proposed. Also, cluster analysis was performed to find out the descriptive features of Promoty’s useful users and based on that, recommendations for improving the referral program were made. The results
are significant for Promoty to achieve fast growth when entering new markets as well as for other startups that are implementing or planning to implement referral programs.
are significant for Promoty to achieve fast growth when entering new markets as well as for other startups that are implementing or planning to implement referral programs.
Graduation Thesis language
Estonian
Graduation Thesis type
Master - Conversion Master in IT
Supervisor(s)
Dr Anna Leontjeva
Defence year
2018