Artificial Intelligence to reduce telecom churn

Artificial Intelligence to reduce telecom churn

Technology News |
By Wisse Hettinga

A major global economic issue in the telecom industry, churn, if not properly managed, leads to significant loss of revenue and blocks the growth of all operators. Economic loss caused by churn is twofold. Firstly, operators lose any future revenue that a churned customer would provide and secondly, all marketing funds used to acquire the customer in the first place are lost.

According to Sicap, a typical mobile operator with an ARPU of €30 in a mature market and subscriber acquisition cost of €270, loses a total of 18% of their annual revenue due to churn, when one assumes an industry average annual churn rate of 24%. The lost subscriber acquisition cost is €65-million per year for one million subscribers, before calculating the lost future revenue.

Sicap AI Engine predicts and identifies churn-prone subscribers, by combining customer-related big data, statistical and analytical techniques and self-learning neuronal networks. The AI Engine makes use of customer data provided by Sicap’s device and SIM management platforms, as well as operators’ other internal and external data sources.

Before the AI Engine is deployed, its neuronal network system is trained by using an operator’s historic data. To increase the prediction accuracy over time, the training is continued using the operator’s actual data.

The system provides a churn prediction list including potential causes for churn and subscriber segments, based on their likelihood to churn within certain confidence intervals. The results are then used to automatically engage customers with targeted and personalized incentive programmes and offers, depending on the segment the subscriber belongs to. Accurate targeting results in more relevant offers, and prevents customers from churning.

The results gathered from the first artificial intelligence-powered churn reduction proof of concepts are convincing. Using predictive and adaptive data models, the subscribers who are likely to churn were identified with 85% precision. Several demographics were identified having higher-than-average churn probability, for example: youngsters, married people, subscribers with a higher call drop rate and more customer care complaints, and those who do not subscribe to additional services.

The identification of the right set of models and parameters for prediction is dependent of the available data. There are numerous modelling techniques for predicting customer churn, which vary in terms of statistical techniques and variable selection methods. Each mobile operator is unique and Sicap’s data analysts will work closely with customers to identify the best possible statistical techniques for each client.

“When properly adapted with a mobile operator’s device, SIM and other data, our predictive churn reduction solution has the potential to save our customers tens of millions of euro annually by targeting the right customers, with the right incentives, at the right moment,” says Markus Doetsch, the CEO of Sicap. “Our aim is to begin with several churn reduction pilot projects with selected operator customers over the next few months.” Doetsch concludes.

If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News


Linked Articles