On the Effects of User Ratings on the Profitability of Cloud Services

In todays cloud market, providers are taking advantage of consumer reviews and ratings as a new marketing tool to establish their credibility. However, to achieve higher ratings, they need to enhance their service quality which comes with an additional cost. In this paper, we model this conflicting situation as a Stackelberg game between a typical service provider and multiple service users in a cloud environment. The strategy of the service provider is to adjust the price and IT capacity by predicting the users ratings as well as their demands variation in response to his given price, quality and rating. The game is solved through a backward induction procedure using Lagrange function and Kuhn-Tucker conditions. To evaluate the proposed model, we performed experiments on three real world service providers who have low, medium and high average of users' ratings, obtained from the Trust Feedback Dataset in the Cloud Armor project. The results show that improvement in ratings is mostly profitable for highly rated providers. The surprising point is that providers having low ratings do not get much benefit from increasing their average ratings, meanwhile, they can perform well when they lower the service price.