Using Big Data To Solve Telcos' Frauds, Revenue leakage

by CXOtoday News Desk    Aug 24, 2017

revenue

Revenue leakage is a real problem faced by telecom operators around the world. Globally about 13% of the telecom revenue is lost due to frauds and revenue leakages (see Figure 1).  Whether it is significant enough for you to pay attention or not is your call, but plugging the gaps should be a concern for every telecom operator.

The revenue assurance program can ensure that these loopholes are identified and plugged to maximize the ARPU and CLV for the business. Fraud, on the other hand, has the capacity to hit telecom businesses with double impact. One you end up losing immediate revenue opportunity, and two, it is possible that some of your customers will be affected and switch.

top fraud

Therefore, fraud detection should be a bigger concern for operators. However, detection and recovery are the things of the past. You need a proactive approach towards revenue assurance.

cxo revenue

RA/FM No More an Isolated Activity

A proactive revenue assurance activity cannot be an isolated activity. It’s no longer an Ops oriented application that can be assigned to a process manager and waited to be resolved. In fact, you will need to consider all the touchpoints in the whole value chain (see Figure 4).

Modern Telecom EcoSystem

Four must have programs for cost effective revenue assurance:

- End user engagement (Consumer CVM)

- Retailer Engagement (Retailer CVM)

- Mobile Money Engagement (Mobile CVM)

- Omni Channel Consumer Care (Digital Care)

- Seamless Purchase Management

These activities are CVM oriented, thus, ensures a better value from the services. To implement these, your big data analytics solution will need the following capabilities:

- Real time streaming data analytics

- Predictive Analytics

- Real time marketing/ promotion capabilities

- Cloud or on-premise deployment capability (could also be a hybrid structure)

- Scalability of the SaaS & Hard infrastructure (if any)

RA/FM with Big Data Analytics

Telecom industries have a huge customer base that is growing fast each passing year. More than 50% of these customers are using smartphones, and even the other 50% who are not, generate a huge amount of data, called user data. This data is usually unstructured and contains secrets to your consumers’ behaviour patterns. The size of this data puzzled the consumer services firms for many years, while they grappled with what to do with it. However, with technology came the big data analytics solutions, offering to solve this puzzle.

Telecom Big Data Analytics

Since the data is generated and centered around the consumers, it can be put to use for consumer facing operations. More than 40% of big data analytics is employed in customer oriented operations. However, that’s not all. This same data also contains secrets to mitigating the risks to your revenue streams.

As shown in figure 4, big data analytics has a role to play in not just customer life time value but also in maintaining service quality, predicting QoE (quality of experience) and countering network frauds.

For example, SIM Box fraud alone is responsible for 3 – 6% of revenue loss for telecom operators globally, and there are two ways to detect it – 1. Make test calls, 2. Passive CDR analysis. While the first approach requires the operator to have a global coverage for the calls, the second approach is rather cheap. The CDR analysis approach is the one that has big data application.

Future of RA/FM with Big Data Analytics Solutions

future

Humans learn and evolve, and that is an ongoing and everlasting challenge for all businesses. But here, we are not addressing only consumption trends but the trends and evolution of fraudsters as well.

The only way to effectively maintain pace with them is that the CIOs are assisted by intelligent machines, which can utilize data lakes (read data oceans) and bring insights, draw patterns, predict emerging patterns, and assist with use case implementation to prevent frauds.

cxo

Machine learning (ML) and artificial intelligence (AI) offer the solution here. Unsupervised machine learning and data lakes can help you identify new and old types of frauds. Continuous analysis of streaming data can point out the outliers more effectively (including cost effectively), identify crime rings and even pre-empt large attacks.

However, machine learning is intended to assist human RA/FM specialists, not replace them. Therefore, the challenge of the expert human resource is here to stay for a while.