This talk will explain, at a business level, how to solve a class of Behavior Analytics problems, such as Product Recommendations for a web site using the Hadoop stack. I will show, step by step, how to take a plain web site and add the personalization capabilities by: extracting (or generating) user purchase history, setting up a Hadoop cluster on Amazon, developing a recommendation algorithm that runs on Hadoop, enhancing web store with recommendation UI, connecting web store with results of analytics processing. Our platform Qubell is used to deploy, manage and tie together a Web site and Hadoop-based analytics platform. We'll touch on - and demonstrate - how to deploy Hadoop on-demand on the cloud, set-up monitoring, start and stop jobs. This talk will be based on "Behavior Analytics Starter Kit" that is coming out later this month. The Kit is open-source, self-contained and self-deployable, so anyone interested to further investigate the solution will be able to do so, free from costs (other than Amazon costs), after the talk.