How to achieve a better retention in Telco using Social Network Analysis

Telecom session

Julien Sauvage (KXEN)
How to achieve a better retention in Telco using Social Network Analysis

In the Telco sector across the world, typical business applications of Social Network Analysis include customer acquisition through Member-Get-Member campaigns, contagious churn prediction and general improvement of churn management and cross and up sell campaign performance. This presentation will focus on the retention area, e.g. mainly 1. the prevention of churn, 2. the monitoring of churn influence and 3. the possible retention strategies. Our customer is an operator in Europe, part of a global Telecommunication group, and has 18M customers. Social Networks of various types were built using voice CDRs (Call Detailed Records) with different weights, aggregation levels and timeframes. A modularity optimization algorithm was then applied to segment each and every network into small communities. A range of social variables was then generated both at the customer and community levels, and used as additional input variables in the churn models. Such variables appeared in the list of the top 20 predictive variables in the model and increased its performance by 50% in the first decile. Moreover, a diffusion process was mimicked, which helped us identify the key churn influencers, and the reasons why they were influential. Finally, Meta-communities were built and helped the customer design the proper offers for retaining both passive and influential churners.