Predict subscriber churn and act on it in real-time

As the saying goes, ‘prevention is better than cure.” Predicting customer churn propensity and acting on it early is extremely important for mobile operators.

The cost of acquiring a new customer is usually five times greater than that of retaining a customer. According to research firm Gartner, Indian mobile operators’ churn rate is between 3.5 percent and 6 percent per month - one of the highest in the Asia-Pacific region.

Growing customer dissatisfaction and churn in telcos are attributed to many reasons: poor network coverage, billing disputes, brand switching or the selection of sub-optimal tariff plans. Customer churn is more conspicuous in the prepaid segment, which accounts for the vast majority of users in emerging markets.

With the emphasis shifting from acquisition to subscriber value generation, operators are focusing on retention programs. Proactive retention programs require the early detection of ‘potential churners’. This can be done only by analysing real-time subscriber behaviour to track any indicative variations. Proactive tracking requires fully automated, real-time technology for churn prediction and management.

Subscriber Behaviour Analysis

Subscribers’ usage behaviour conveys a lot of information about their mindset, preference and spending patterns. Any variation is an indication of churn tendency.



For example: in general, an average subscriber exhibits a normal variance of 20 percent in calling pattern over a week. If a subscriber does not make any calls for a week, this behaviour is outside the norm. Is this a potential problem, or an opportunity? Does the subscriber need a different calling package? Is the subscriber about to churn? In either case, there is an opportunity to increase the value of the subscriber, if you have the ability to recognise the threat and react to it in time.

In this example, the weekly variance in calling pattern serves as the ‘trip wire’ and suggests the raising of the hand by the subscriber, who is saying to the marketer : ‘I am changing my behaviour. Pay attention to me.’ Now it is upto the marketer to determine the best action. So marketers need to set tripwires like this to detect indicative variation in usage behaviour and predict the possibility of customer dissatisfaction and defection.

Neon’s Real-Time Churn Prediction Engine is designed to constantly analyse the real-time usage profile of all subscribers. All a marketer needs to do is set tripwires to detect behavioural changes - and the right corrective measures will be taken automatically.

With Neon, marketers can define churn detection rules and implement trip wires based on any combination of usage criteria. These criteria include:
Total usage and number of calls (incoming or outgoing)

• Leg-wise calls made (local, STD, ISD)
• Recharge frequency and average recharge amount
• Value-added service usage, number of downloads, etc
• Average revenue per user (ARPU) variations, even at a granular, individual service level