Worldwide CSPs are waking up to the fact that to remain profitable in the face of fierce competition it would be as much important to arrest revenue erosion as it would be to acquire new customers. Churn has emerged as one of the most critical source of revenue erosion. It is a generally held view in the industry that CSPs lose about 40% customers annually. A study in the US revealed that almost 9.9% revenue is lost by CSPs due to customer churn alone. Juniper Research also predicts that $4bn will be lost by CSPs worldwide by 2018 owing to this problem. In developing market it is a much larger menace to an extent of 100% churn every year. It is clearly an issue that needs to be addressed by CSPs with utmost urgency.

Voluntary Vs. Involuntary Churn 

Churn has its root causes attributable to multiple factors. It is almost like chronic diseases which cannot be totally eradicated at one go, but can definitely be controlled and prevented if it can be detected at an early stage. There is voluntary and involuntary churn. Involuntary churn can be due to non-payment of bills, migration and so on. Certainly it does not make sense to try to revive customers who have a bad payment history.

However, on the contrary voluntary churn can be controlled to a great extent. In the course of their interaction with the Operator, at times a customer faces many issues such as inaccurate billing (more popularly known as bill shock in telecom parlance), unsatisfactory resolution of issues by the customer care, lack of connectivity in places the customer frequents, competitive offers, and so on. These can force customers to explore other options and ultimately defection. If you look at these factors closely, they are controllable and within the realm of CSP.

Track the signs of churn early

It is clear that CSPs can definitely control voluntary churn, provided they can detect it early by tracking and analyzing the usage behavior closely. The most common indicators for churn are continually declining monthly usage, reduction or complete withdrawal in spent on additional services, using the connection only for incoming calls, extended period of inactivity and so on. For prepaid subscribers an additional indicator generally observed is infrequent recharging and recharging successively with lesser denomination.

For post-paid connections, signs could be different like subscriber downgrading his plan near the end of the contract period or even opting to switch to prepaid mode. Sometime dissatisfaction can be detected through the increase in the frequency of calls to customer care.

Churn indicators could vary by customer segment

If you dive deeper into historic usage patterns, one could even find that churn indicators vary from one customer segment to the other. For example, a high value customer may not waste too much time calling the customer care with his grievances. If one call does not solve his problem, he would move on. Moreover, high value customers do not like to part with their number, so, they probably would look to port their number. Applying for MNP, then, is a clear indication but it may be too late for a win back. On the other hand, in a prepaid scenario, an extended period of ‘zero usage’ cannot be always concluded as churn behavior as some of them tend to come back on network after a period of inactivity. In such cases, only a detailed study of their usage history can enable marketers to confirm any anomaly of behavior.

Also, it is important to detect if the churn is involuntary, in that case it may not make good business sense to invest time or money on a win back effort. Like such, there could be multitude of factors and indicators across numerous segments that are too complex and too many for human expertise alone to detect and act. It has to be a man-machine collaborative effort!

Big Data Analytics can ‘predict’ and ‘prescribe’ to mitigate churn

Effectiveness of any churn mitigation program boils down to how quickly and accurately marketers can detect signs of churn among the various segments. A vast amount of customer data is generated across various sources like CRM, billing, network, customer touch points and so on. However, data generated from these sources need to be integrated and transformed to easily interpretable and actionable insights so that it gives a holistic picture of the customers’ behavior. Big Data technology has ushered in a new era of innovative new statistical models that can auto-detect these early churn symptoms across different segments and accurately recommend actions to mitigate the churn. A well-designed and robust churn management system not only will predict churn to a higher degree of accuracy but also will enable detailed causal analysis and provide marketers with next best actions to mitigate it on time.