Data Mining & Machine Learning to empower business strategy

Опубликовано 6 сентября 2016, 5:07

This talk will focus on my recent work for the Acquisition and Retention team at MSN, in which we address the business problem of successfully predicting customer behavior far enough ahead that the predictions are actionable.  The goals of this work were; to impact churn, by allowing intervention to retain valuable customers; to allow the monetization of customers leaving the MSN dialup service for Broadband, by targeting high upgrade propensity customers to partners, and focusing those living outside areas with a partner relationship on MSN’s BYOA (Premium) product. Using the SQL Server 2005 data mining tools, we developed and deployed a set of 4 predictive models on MSN’s database of nearly 5 million active subscribers, around which MSN’s Acquisition & Retention team now builds business strategy. During the course of this work, we uncovered a simple abstraction of the subscription relationship that can be applied to any contract-free subscription business, in particular the post-number-portability wireless industry.