Telcos are sitting on a mountain of data. It pours in like a river from customer interactions, their bills and their usage.
Telcos are investing billions of dollars bringing together customer data and gathering more customer data from a variety of sources. New and easier reporting brings access to this data and new tools are making the tasks of predictive churn models easier to produce.
However, in most cases the models are virtually the same as those constructed years ago, based on outdated segmentation and inaccurate assumptions. Telcos need to examine the data that they have (or have access to) and how it can be used to manage churn. Retention strategies should be tailored based on customer segment, lifetime value and buyer preferences.
Get the basics right Multiple interaction points from traditional call centers and more recent web and mobile interfaces provide operators with a wealth of data points and transaction detail. Hidden in the data is the key to the next generation of offerings as well as "predicting the prediction". Operators can now determine not only the customer's social and professional segmentation through static data (traditional approach) but also more dynamic characteristics, such as customer locale (internationally through roaming), entertainment preferences, shopping patterns and television viewing behaviors.
Combining this information allows for true customer insight into the buyer's needs and values. A couple of years ago, an APAC triple-play operator determined the correlation between mobile plan usage and TV behaviors through advanced analytics and segmentation. It then tailored its cross-selling strategies based on the identified correlation. Naturally, telcos still need to be cognizant of privacy concerns. However, there are less customers' complains regarding privacy as Facebook and Google have already set the stage for a privacy "gray zone".
Some telcos are using customer knowledge and advanced analytics capabilities to manage their churn through predictive modeling. Sophisticated churn modeling techniques have long been used to predict different churn behaviors. For example, churn models are used to predict churn along key points of the customer lifecycle. The most obvious example is increasing churn as customers are getting closer to contract end date. Churn models are also combined with churn trigger events to make retention efforts more relevant. For instance, research shows that following a single poor customer experience event, a customer will churn 80% of the time during the next opportunity.